<x:xmpmeta tsc:version="2.0.1" xmlns:x="adobe:ns:meta/" xmlns:tsc="http://www.techsmith.com/xmp/tsc/">
   <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:xmp="http://ns.adobe.com/xap/1.0/" xmlns:xmpDM="http://ns.adobe.com/xmp/1.0/DynamicMedia/" xmlns:xmpG="http://ns.adobe.com/xap/1.0/g/" xmlns:xmpMM="http://ns.adobe.com/xap/1.0/mm/" xmlns:tscDM="http://www.techsmith.com/xmp/tscDM/" xmlns:tscIQ="http://www.techsmith.com/xmp/tscIQ/" xmlns:tscHS="http://www.techsmith.com/xmp/tscHS/" xmlns:stDim="http://ns.adobe.com/xap/1.0/sType/Dimensions#" xmlns:stFnt="http://ns.adobe.com/xap/1.0/sType/Font#" xmlns:exif="http://ns.adobe.com/exif/1.0" xmlns:dc="http://purl.org/dc/elements/1.1/">
      <rdf:Description dc:date="2017-09-28 10:13:23 PM" dc:source="Camtasia,9.0.5,enu" dc:title="k_15orm2" tscDM:firstFrame="k_15orm2_First_Frame.png" tscDM:originId="2625EA67-814F-4A0C-B4C7-2902396CFF3B" tscDM:project="k_15orm2.tscproj">
         <xmpDM:duration xmpDM:scale="1/1000" xmpDM:value="3554600"/>
         <xmpDM:videoFrameSize stDim:unit="pixel" stDim:h="606" stDim:w="810"/>
         <tsc:langName>
            <rdf:Bag>
               <rdf:li xml:lang="en-US">English</rdf:li></rdf:Bag>
         </tsc:langName>
         <xmpDM:Tracks>
            <rdf:Bag>
               <rdf:li>
                  <rdf:Description xmpDM:trackType="ScreenText" xmpDM:frameRate="f1000" xmpDM:trackName="Screen Text">
                     <xmpDM:markers>
                        <rdf:Seq>
                           <rdf:li><rdf:Description xmpDM:name="Problem of Multicollinearity Two lines are said to be collinear if they have the same slope and same intercept." xmpDM:startTime="0" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="These two lines can be represented in just one dimension instead of the two dimensions required for lines which are" xmpDM:startTime="0" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="not collinear." xmpDM:startTime="0" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Two collinear lines are not independent." xmpDM:startTime="0" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="When two predictor variables are linearly dependent, they are called collinear." xmpDM:startTime="0" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Collinear predictors Þ Problem of multicollinearity Þ Contradictory results from various significance tests." xmpDM:startTime="0" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="High Correlation Þ Eliminate one variable and check if significance improves" xmpDM:startTime="0" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.2" xmpDM:startTime="58833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For the Disk-Memory-CPU data of Example15.1" xmpDM:startTime="58833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Computed F ratio &gt; F value from the table Þ Regression does explain a significant part of the variation Note:" xmpDM:startTime="58833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Regression passed the F test Þ Hypothesis of all parameters being zero cannot be accepted." xmpDM:startTime="58833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="However, none of the regression parameters are significantly different from zero." xmpDM:startTime="58833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="This contradiction Þ Problem of multicollinearity TP_tmp" xmpDM:startTime="58833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Problem of Multicollinearity Two lines are said to be collinear if they have the same slope and same intercept." xmpDM:startTime="83433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="These two lines can be represented in just one dimension instead of the two dimensions required for lines which are" xmpDM:startTime="83433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="not collinear." xmpDM:startTime="83433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Two collinear lines are not independent." xmpDM:startTime="83433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="When two predictor variables are linearly dependent, they are called collinear." xmpDM:startTime="83433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Collinear predictors Þ Problem of multicollinearity Þ Contradictory results from various significance tests." xmpDM:startTime="83433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="High Correlation Þ Eliminate one variable and check if significance improves" xmpDM:startTime="83433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3" xmpDM:startTime="96367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For the data of Example 15.2," xmpDM:startTime="96367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="n=7, S x1i=271, S x2i=1324, S x1i2=1385, S x2i2=326,686, S x1ix2i=67,188." xmpDM:startTime="96367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Correlation is high Þ Programs with large memory sizes have more I/O's In Example14.1," xmpDM:startTime="96367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="CPU time on number of disk I/O's regression was found significant." xmpDM:startTime="96367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="96367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3" xmpDM:startTime="173333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Similarly, in Exercise 14.3," xmpDM:startTime="173333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="CPU time is regressed on the memory size and the resulting regression parameters are found to be significant." xmpDM:startTime="173333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Thus, either the number of I/O's or the memory size can be used to estimate CPU time, but not both." xmpDM:startTime="173333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Lesson: Adding a predictor variable does not always improve a regression." xmpDM:startTime="173333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If the variable is correlated to other predictors, it may reduce the statistical accuracy of the regression." xmpDM:startTime="173333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Try all 2k possible subsets and choose the one that gives the best results with small number of variables." xmpDM:startTime="173333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Correlation matrix for the subset chosen should be checked" xmpDM:startTime="173333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Exercise 15.1" xmpDM:startTime="278233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Updated) The results of a multiple regression based on 9 observations are shown in the following table." xmpDM:startTime="278233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp Standard deviation of errors = 12.0" xmpDM:startTime="278233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Intercept=75.3" xmpDM:startTime="278233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="278233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont) Based on these results answer the following questions (Show computation to justify your answers): What percent of" xmpDM:startTime="315600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="variation is explained by the regression? Is the regression significant at 90% confidence level? Which variable has the highest coefficient?" xmpDM:startTime="315600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(bj is the coefficient of xj) Which variable is most significant? Which coefficients are significant at 90%? What is the" xmpDM:startTime="315600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="problem with this regression? What would you try next?" xmpDM:startTime="315600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Exercise 15.1" xmpDM:startTime="321467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Updated) The results of a multiple regression based on 9 observations are shown in the following table." xmpDM:startTime="321467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp Standard deviation of errors = 12.0" xmpDM:startTime="321467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Intercept=75.3" xmpDM:startTime="321467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="321467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont) Based on these results answer the following questions (Show computation to justify your answers): What percent of" xmpDM:startTime="368167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="variation is explained by the regression? Is the regression significant at 90% confidence level? Which variable has the highest coefficient?" xmpDM:startTime="368167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(bj is the coefficient of xj) Which variable is most significant? Which coefficients are significant at 90%? What is the" xmpDM:startTime="368167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="problem with this regression? What would you try next?" xmpDM:startTime="368167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Exercise 15.1" xmpDM:startTime="385300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Updated) The results of a multiple regression based on 9 observations are shown in the following table." xmpDM:startTime="385300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp Standard deviation of errors = 12.0" xmpDM:startTime="385300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Intercept=75.3" xmpDM:startTime="385300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="385300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont) Based on these results answer the following questions (Show computation to justify your answers): What percent of" xmpDM:startTime="416667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="variation is explained by the regression? Is the regression significant at 90% confidence level? Which variable has the highest coefficient?" xmpDM:startTime="416667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(bj is the coefficient of xj) Which variable is most significant? Which coefficients are significant at 90%? What is the" xmpDM:startTime="416667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="problem with this regression? What would you try next?" xmpDM:startTime="416667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Exercise 15.1" xmpDM:startTime="419967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Updated) The results of a multiple regression based on 9 observations are shown in the following table." xmpDM:startTime="419967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp Standard deviation of errors = 12.0" xmpDM:startTime="419967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Intercept=75.3" xmpDM:startTime="419967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="419967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3" xmpDM:startTime="431100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Similarly, in Exercise 14.3," xmpDM:startTime="431100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="CPU time is regressed on the memory size and the resulting regression parameters are found to be significant." xmpDM:startTime="431100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Thus, either the number of I/O's or the memory size can be used to estimate CPU time, but not both." xmpDM:startTime="431100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Lesson: Adding a predictor variable does not always improve a regression." xmpDM:startTime="431100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If the variable is correlated to other predictors, it may reduce the statistical accuracy of the regression." xmpDM:startTime="431100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Try all 2k possible subsets and choose the one that gives the best results with small number of variables." xmpDM:startTime="431100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Correlation matrix for the subset chosen should be checked" xmpDM:startTime="431100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3" xmpDM:startTime="431267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For the data of Example 15.2," xmpDM:startTime="431267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="n=7, S x1i=271, S x2i=1324, S x1i2=1385, S x2i2=326,686, S x1ix2i=67,188." xmpDM:startTime="431267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Correlation is high Þ Programs with large memory sizes have more I/O's In Example14.1," xmpDM:startTime="431267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="CPU time on number of disk I/O's regression was found significant." xmpDM:startTime="431267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="431267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Problem of Multicollinearity Two lines are said to be collinear if they have the same slope and same intercept." xmpDM:startTime="431400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="These two lines can be represented in just one dimension instead of the two dimensions required for lines which are" xmpDM:startTime="431400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="not collinear." xmpDM:startTime="431400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Two collinear lines are not independent." xmpDM:startTime="431400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="When two predictor variables are linearly dependent, they are called collinear." xmpDM:startTime="431400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Collinear predictors Þ Problem of multicollinearity Þ Contradictory results from various significance tests." xmpDM:startTime="431400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="High Correlation Þ Eliminate one variable and check if significance improves" xmpDM:startTime="431400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.2" xmpDM:startTime="431567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For the Disk-Memory-CPU data of Example15.1" xmpDM:startTime="431567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Computed F ratio &gt; F value from the table Þ Regression does explain a significant part of the variation Note:" xmpDM:startTime="431567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Regression passed the F test Þ Hypothesis of all parameters being zero cannot be accepted." xmpDM:startTime="431567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="However, none of the regression parameters are significantly different from zero." xmpDM:startTime="431567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="This contradiction Þ Problem of multicollinearity TP_tmp" xmpDM:startTime="431567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ANOVA Table for Multiple Linear Regression Table 15.3" xmpDM:startTime="431733" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="431733" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="F-Test (Cont) MSE = Variance of Error MSR/MSE has F[k, n-k-1] distribution F-test = Null hypothesis that y doesn't depend" xmpDM:startTime="431900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="upon any xj: against an alternate hypothesis that y depends upon at least one xj and therefore, at least one" xmpDM:startTime="431900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="bj ¹ 0." xmpDM:startTime="431900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If the computed ratio is less than the value read from the table, the null hypothesis cannot be rejected at" xmpDM:startTime="431900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="the stated significance level." xmpDM:startTime="431900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="In simple regression models, If the confidence interval of b1 does not include zero Þ Parameter is nonzero Þ Regression" xmpDM:startTime="431900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="explains a significant part of the response variation Þ F-test is not required." xmpDM:startTime="431900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp and" xmpDM:startTime="431900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="F-Test Given SSi and SSj with ni and nj degrees of freedom, the ratio (SSi/ni)/(SSj/nj) has an F distribution with" xmpDM:startTime="432100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ni numerator degrees of freedom and nj denominator degrees of freedom." xmpDM:startTime="432100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Hypothesis that the sum SSi is less than or equal to SSj is rejected at a significance level, if the" xmpDM:startTime="432100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ratio (SSi/ni)/(SSj/nj) is greater than the 1-a quantile of the F-variate." xmpDM:startTime="432100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="This procedure is also known as F-test." xmpDM:startTime="432100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The F-test can be used to check: Is SSR is significantly higher than SSE? Þ Use F-test Þ Compute (SSR/nR)/(SSE/ne)" xmpDM:startTime="432100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="= MSR/MSE" xmpDM:startTime="432100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Analysis of Variance Test the hypothesis that SSR is less than or equal to SSE." xmpDM:startTime="432300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Degrees of freedom = Number of independent values required to compute Assuming Errors are i.i.d." xmpDM:startTime="432300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Normal Þ y's are also normally distributed, x's are nonstochastic Þ Can be measured without errors Various sums of squares" xmpDM:startTime="432300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="have a chi-square distribution with the degrees of freedom as given above." xmpDM:startTime="432300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="432300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1" xmpDM:startTime="432500" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Standard deviation for a mean of large number of observations is: 90% confidence interval is: TP_tmp TP_tmp" xmpDM:startTime="432500" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1" xmpDM:startTime="432700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) A single future observation for programs with 100 disk I/O's and a memory size of 550: Standard deviation of" xmpDM:startTime="432700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="the predicted observation is: 90% confidence interval using the t value of 2.132" xmpDM:startTime="432700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="is: TP_tmp TP_tmp TP_tmp" xmpDM:startTime="432700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1" xmpDM:startTime="433033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Standard deviations of the regression parameters are: The 0.95-quantile" xmpDM:startTime="433033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="for a t-variate with four degrees of freedom is 2.132." xmpDM:startTime="433033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="None of the three parameters is significant at a 90% confidence level." xmpDM:startTime="433033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="433033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1" xmpDM:startTime="433433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) The coefficient of determination R2 is: Thus, the regression explains 97% of the variation of y." xmpDM:startTime="433433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Coefficient of multiple correlation: Standard deviation of errors is: TP_tmp TP_tmp TP_tmp" xmpDM:startTime="433433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1" xmpDM:startTime="465533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Standard deviations of the regression parameters are: The 0.95-quantile" xmpDM:startTime="465533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="for a t-variate with four degrees of freedom is 2.132." xmpDM:startTime="465533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="None of the three parameters is significant at a 90% confidence level." xmpDM:startTime="465533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="465533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1" xmpDM:startTime="465800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) A single future observation for programs with 100 disk I/O's and a memory size of 550: Standard deviation of" xmpDM:startTime="465800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="the predicted observation is: 90% confidence interval using the t value of 2.132" xmpDM:startTime="465800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="is: TP_tmp TP_tmp TP_tmp" xmpDM:startTime="465800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1" xmpDM:startTime="466033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Standard deviation for a mean of large number of observations is: 90% confidence interval is: TP_tmp TP_tmp" xmpDM:startTime="466033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Analysis of Variance Test the hypothesis that SSR is less than or equal to SSE." xmpDM:startTime="466400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Degrees of freedom = Number of independent values required to compute Assuming Errors are i.i.d." xmpDM:startTime="466400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Normal Þ y's are also normally distributed, x's are nonstochastic Þ Can be measured without errors Various sums of squares" xmpDM:startTime="466400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="have a chi-square distribution with the degrees of freedom as given above." xmpDM:startTime="466400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="466400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="F-Test Given SSi and SSj with ni and nj degrees of freedom, the ratio (SSi/ni)/(SSj/nj) has an F distribution with" xmpDM:startTime="466833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ni numerator degrees of freedom and nj denominator degrees of freedom." xmpDM:startTime="466833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Hypothesis that the sum SSi is less than or equal to SSj is rejected at a significance level, if the" xmpDM:startTime="466833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ratio (SSi/ni)/(SSj/nj) is greater than the 1-a quantile of the F-variate." xmpDM:startTime="466833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="This procedure is also known as F-test." xmpDM:startTime="466833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The F-test can be used to check: Is SSR is significantly higher than SSE? Þ Use F-test Þ Compute (SSR/nR)/(SSE/ne)" xmpDM:startTime="466833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="= MSR/MSE" xmpDM:startTime="466833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="F-Test (Cont) MSE = Variance of Error MSR/MSE has F[k, n-k-1] distribution F-test = Null hypothesis that y doesn't depend" xmpDM:startTime="467233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="upon any xj: against an alternate hypothesis that y depends upon at least one xj and therefore, at least one" xmpDM:startTime="467233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="bj ¹ 0." xmpDM:startTime="467233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If the computed ratio is less than the value read from the table, the null hypothesis cannot be rejected at" xmpDM:startTime="467233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="the stated significance level." xmpDM:startTime="467233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="In simple regression models, If the confidence interval of b1 does not include zero Þ Parameter is nonzero Þ Regression" xmpDM:startTime="467233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="explains a significant part of the response variation Þ F-test is not required." xmpDM:startTime="467233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp and" xmpDM:startTime="467233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ANOVA Table for Multiple Linear Regression Table 15.3" xmpDM:startTime="467633" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="467633" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.2" xmpDM:startTime="468133" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For the Disk-Memory-CPU data of Example15.1" xmpDM:startTime="468133" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Computed F ratio &gt; F value from the table Þ Regression does explain a significant part of the variation Note:" xmpDM:startTime="468133" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Regression passed the F test Þ Hypothesis of all parameters being zero cannot be accepted." xmpDM:startTime="468133" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="However, none of the regression parameters are significantly different from zero." xmpDM:startTime="468133" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="This contradiction Þ Problem of multicollinearity TP_tmp" xmpDM:startTime="468133" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Problem of Multicollinearity Two lines are said to be collinear if they have the same slope and same intercept." xmpDM:startTime="468533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="These two lines can be represented in just one dimension instead of the two dimensions required for lines which are" xmpDM:startTime="468533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="not collinear." xmpDM:startTime="468533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Two collinear lines are not independent." xmpDM:startTime="468533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="When two predictor variables are linearly dependent, they are called collinear." xmpDM:startTime="468533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Collinear predictors Þ Problem of multicollinearity Þ Contradictory results from various significance tests." xmpDM:startTime="468533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="High Correlation Þ Eliminate one variable and check if significance improves" xmpDM:startTime="468533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3" xmpDM:startTime="469067" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For the data of Example 15.2," xmpDM:startTime="469067" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="n=7, S x1i=271, S x2i=1324, S x1i2=1385, S x2i2=326,686, S x1ix2i=67,188." xmpDM:startTime="469067" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Correlation is high Þ Programs with large memory sizes have more I/O's In Example14.1," xmpDM:startTime="469067" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="CPU time on number of disk I/O's regression was found significant." xmpDM:startTime="469067" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="469067" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3" xmpDM:startTime="469567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Similarly, in Exercise 14.3," xmpDM:startTime="469567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="CPU time is regressed on the memory size and the resulting regression parameters are found to be significant." xmpDM:startTime="469567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Thus, either the number of I/O's or the memory size can be used to estimate CPU time, but not both." xmpDM:startTime="469567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Lesson: Adding a predictor variable does not always improve a regression." xmpDM:startTime="469567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If the variable is correlated to other predictors, it may reduce the statistical accuracy of the regression." xmpDM:startTime="469567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Try all 2k possible subsets and choose the one that gives the best results with small number of variables." xmpDM:startTime="469567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Correlation matrix for the subset chosen should be checked" xmpDM:startTime="469567" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Exercise 15.1" xmpDM:startTime="470033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Updated) The results of a multiple regression based on 9 observations are shown in the following table." xmpDM:startTime="470033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp Standard deviation of errors = 12.0" xmpDM:startTime="470033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Intercept=75.3" xmpDM:startTime="470033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="470033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont) Based on these results answer the following questions (Show computation to justify your answers): What percent of" xmpDM:startTime="471167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="variation is explained by the regression? Is the regression significant at 90% confidence level? Which variable has the highest coefficient?" xmpDM:startTime="471167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(bj is the coefficient of xj) Which variable is most significant? Which coefficients are significant at 90%? What is the" xmpDM:startTime="471167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="problem with this regression? What would you try next?" xmpDM:startTime="471167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Exercise 15.1" xmpDM:startTime="513100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Updated) The results of a multiple regression based on 9 observations are shown in the following table." xmpDM:startTime="513100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp Standard deviation of errors = 12.0" xmpDM:startTime="513100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Intercept=75.3" xmpDM:startTime="513100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="513100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont) Based on these results answer the following questions (Show computation to justify your answers): What percent of" xmpDM:startTime="518233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="variation is explained by the regression? Is the regression significant at 90% confidence level? Which variable has the highest coefficient?" xmpDM:startTime="518233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(bj is the coefficient of xj) Which variable is most significant? Which coefficients are significant at 90%? What is the" xmpDM:startTime="518233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="problem with this regression? What would you try next?" xmpDM:startTime="518233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Regression with Categorical Predictors Note: If all predictor variables are categorical, use one of the experimental design and analysis techniques" xmpDM:startTime="572800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="for statistically more precise (less variant) results Use regression if most predictors are quantitative and only a few predictors are" xmpDM:startTime="572800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="categorical." xmpDM:startTime="572800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Two Categories: bj = difference in the effect of the two alternatives bj = Insignificant Þ Two alternatives have similar" xmpDM:startTime="572800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="performance Alternatively: bj = Difference from the average response Difference of the effects of the two levels is 2bj TP_tmp" xmpDM:startTime="572800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="572800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Three Categories: Incorrect: This coding implies an order Þ B is half way between A and C." xmpDM:startTime="727600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="This may not be true." xmpDM:startTime="727600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Recommended: Use two predictor variables TP_tmp TP_tmp" xmpDM:startTime="727600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Thus, This coding does not imply any ordering among the types." xmpDM:startTime="777967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Provides an easy way to interpret the regression parameters." xmpDM:startTime="777967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="777967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) The average responses for the three types are: Thus, b1 represents the difference between type A and" xmpDM:startTime="810933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="C." xmpDM:startTime="810933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b2 represents the difference between type B and C." xmpDM:startTime="810933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 represents type C." xmpDM:startTime="810933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="810933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Thus, This coding does not imply any ordering among the types." xmpDM:startTime="817867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Provides an easy way to interpret the regression parameters." xmpDM:startTime="817867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="817867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) The average responses for the three types are: Thus, b1 represents the difference between type A and" xmpDM:startTime="825700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="C." xmpDM:startTime="825700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b2 represents the difference between type B and C." xmpDM:startTime="825700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 represents type C." xmpDM:startTime="825700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="825700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Thus, This coding does not imply any ordering among the types." xmpDM:startTime="837433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Provides an easy way to interpret the regression parameters." xmpDM:startTime="837433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="837433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) The average responses for the three types are: Thus, b1 represents the difference between type A and" xmpDM:startTime="856533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="C." xmpDM:startTime="856533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b2 represents the difference between type B and C." xmpDM:startTime="856533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 represents type C." xmpDM:startTime="856533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="856533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Level = Number of values that a categorical variable can take To represent a categorical variable with" xmpDM:startTime="858000" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="k levels, define k-1 binary variables: kth (last) value is defined by x1= x2= L = xk-1= 0." xmpDM:startTime="858000" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 = Average response with the kth alternative." xmpDM:startTime="858000" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="bj = Difference between alternatives j and k." xmpDM:startTime="858000" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If one of the alternatives represents the status quo or a standard against which other alternatives have to be measured," xmpDM:startTime="858000" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="that alternative should be coded as the kth alternative." xmpDM:startTime="858000" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="858000" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) The average responses for the three types are: Thus, b1 represents the difference between type A and" xmpDM:startTime="874933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="C." xmpDM:startTime="874933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b2 represents the difference between type B and C." xmpDM:startTime="874933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 represents type C." xmpDM:startTime="874933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="874933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Thus, This coding does not imply any ordering among the types." xmpDM:startTime="924767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Provides an easy way to interpret the regression parameters." xmpDM:startTime="924767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="924767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Three Categories: Incorrect: This coding implies an order Þ B is half way between A and C." xmpDM:startTime="928167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="This may not be true." xmpDM:startTime="928167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Recommended: Use two predictor variables TP_tmp TP_tmp" xmpDM:startTime="928167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Thus, This coding does not imply any ordering among the types." xmpDM:startTime="945167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Provides an easy way to interpret the regression parameters." xmpDM:startTime="945167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="945167" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) The average responses for the three types are: Thus, b1 represents the difference between type A and" xmpDM:startTime="948700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="C." xmpDM:startTime="948700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b2 represents the difference between type B and C." xmpDM:startTime="948700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 represents type C." xmpDM:startTime="948700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="948700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Level = Number of values that a categorical variable can take To represent a categorical variable with" xmpDM:startTime="950667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="k levels, define k-1 binary variables: kth (last) value is defined by x1= x2= L = xk-1= 0." xmpDM:startTime="950667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 = Average response with the kth alternative." xmpDM:startTime="950667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="bj = Difference between alternatives j and k." xmpDM:startTime="950667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If one of the alternatives represents the status quo or a standard against which other alternatives have to be measured," xmpDM:startTime="950667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="that alternative should be coded as the kth alternative." xmpDM:startTime="950667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="950667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) The average responses for the three types are: Thus, b1 represents the difference between type A and" xmpDM:startTime="1038967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="C." xmpDM:startTime="1038967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b2 represents the difference between type B and C." xmpDM:startTime="1038967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 represents type C." xmpDM:startTime="1038967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="1038967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Level = Number of values that a categorical variable can take To represent a categorical variable with" xmpDM:startTime="1069467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="k levels, define k-1 binary variables: kth (last) value is defined by x1= x2= L = xk-1= 0." xmpDM:startTime="1069467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 = Average response with the kth alternative." xmpDM:startTime="1069467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="bj = Difference between alternatives j and k." xmpDM:startTime="1069467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If one of the alternatives represents the status quo or a standard against which other alternatives have to be measured," xmpDM:startTime="1069467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="that alternative should be coded as the kth alternative." xmpDM:startTime="1069467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="1069467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1:" xmpDM:startTime="1070100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="RPC performance RPC performance on Unix and Argus where, y is the elapsed time, x1 is the data size and" xmpDM:startTime="1070100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp" xmpDM:startTime="1070100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1" xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) All three parameters are significant." xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The regression explains 76.5%" xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="of the variation." xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Per byte processing cost (time) for both operating systems is 0.025" xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="millisecond." xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Set up cost is 36.73" xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="milliseconds on Argus which is 14.927" xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="milliseconds more than that with Unix." xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="1070533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1:" xmpDM:startTime="1071367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="RPC performance RPC performance on Unix and Argus where, y is the elapsed time, x1 is the data size and" xmpDM:startTime="1071367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp" xmpDM:startTime="1071367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Categorical Predictors (Cont) Level = Number of values that a categorical variable can take To represent a categorical variable with" xmpDM:startTime="1072667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="k levels, define k-1 binary variables: kth (last) value is defined by x1= x2= L = xk-1= 0." xmpDM:startTime="1072667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="b0 = Average response with the kth alternative." xmpDM:startTime="1072667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="bj = Difference between alternatives j and k." xmpDM:startTime="1072667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If one of the alternatives represents the status quo or a standard against which other alternatives have to be measured," xmpDM:startTime="1072667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="that alternative should be coded as the kth alternative." xmpDM:startTime="1072667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="1072667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1:" xmpDM:startTime="1074233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="RPC performance RPC performance on Unix and Argus where, y is the elapsed time, x1 is the data size and" xmpDM:startTime="1074233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp" xmpDM:startTime="1074233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1" xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) All three parameters are significant." xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The regression explains 76.5%" xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="of the variation." xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Per byte processing cost (time) for both operating systems is 0.025" xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="millisecond." xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Set up cost is 36.73" xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="milliseconds on Argus which is 14.927" xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="milliseconds more than that with Unix." xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="1145867" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1:" xmpDM:startTime="1270533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="RPC performance RPC performance on Unix and Argus where, y is the elapsed time, x1 is the data size and" xmpDM:startTime="1270533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp" xmpDM:startTime="1270533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1" xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) All three parameters are significant." xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The regression explains 76.5%" xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="of the variation." xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Per byte processing cost (time) for both operating systems is 0.025" xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="millisecond." xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Set up cost is 36.73" xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="milliseconds on Argus which is 14.927" xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="milliseconds more than that with Unix." xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="1290533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1:" xmpDM:startTime="1310033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="RPC performance RPC performance on Unix and Argus where, y is the elapsed time, x1 is the data size and" xmpDM:startTime="1310033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp" xmpDM:startTime="1310033" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1" xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) All three parameters are significant." xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The regression explains 76.5%" xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="of the variation." xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Per byte processing cost (time) for both operating systems is 0.025" xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="millisecond." xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Set up cost is 36.73" xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="milliseconds on Argus which is 14.927" xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="milliseconds more than that with Unix." xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="1347100" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Differing Conclusions Case Study 14.1" xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="concluded that there was no significant difference in the set up costs." xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The per byte costs were different." xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1" xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="concluded that per byte cost is same but the set up costs are different." xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Which conclusion is correct? Need system (domain) knowledge." xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Statistical techniques applied without understanding the system can lead to a misleading result." xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 14.1" xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="was based on the assumption that the processing as well as set up in the two operating systems are different" xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Þ Four parameters The data showed that the setup costs were numerically indistinguishable." xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="1350467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Differing Conclusions (Cont) The model used in Case Study 15.1" xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="is based on the assumption that the operating systems have no effect on per byte processing." xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="This will be true if the processing is identical on the two systems and does not involve the operating systems." xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Only set up requires operating system calls." xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If this is, in fact, true then the regression coefficients estimated in the joint model of this case study 15.1" xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="are more realistic estimates of the real world." xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="On the other hand, if system programmers can show that the processing follows a different code path in the two" xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="systems, then the model of Case Study 14.1" xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="would be more realistic." xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="1502433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Curvilinear Regression If the relationship between response and predictors is nonlinear but it can be converted into a linear form" xmpDM:startTime="1540400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Þ curvilinear regression." xmpDM:startTime="1540400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example: Taking a logarithm of both sides we get: Thus, ln x and ln y are linearly related." xmpDM:startTime="1540400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The values of ln b and a can be found by a linear regression of ln y on ln x." xmpDM:startTime="1540400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="1540400" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Curvilinear Regression: Other Examples If a predictor variable appears in more than one transformed predictor variables, the transformed variables are" xmpDM:startTime="1644300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="likely to be correlated Þ multicollinearity." xmpDM:startTime="1644300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Try various possible subsets of the predictor variables to find a subset that gives significant parameters and explains a high" xmpDM:startTime="1644300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="percentage of the observed variation." xmpDM:startTime="1644300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="1644300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="1795333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Amdahl's law: I/O rate is proportional to the processor speed." xmpDM:startTime="1795333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For each instruction executed there is one bit of I/O on the average." xmpDM:startTime="1795333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="1795333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="1883233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Let us fit the following curvilinear model to this data: Taking a log of both sides we get: TP_tmp" xmpDM:startTime="1883233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="1883233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="1930967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Both coefficients are significant at 90% confidence level." xmpDM:startTime="1930967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The regression explains 84% of the variation." xmpDM:startTime="1930967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="At this confidence level, we can accept the hypothesis that the relationship is linear since the confidence interval for b1" xmpDM:startTime="1930967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="includes 1." xmpDM:startTime="1930967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="1930967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2012700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Errors in log I/O rate do seem to be normally distributed." xmpDM:startTime="2012700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TPps2bmp" xmpDM:startTime="2012700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2036700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Both coefficients are significant at 90% confidence level." xmpDM:startTime="2036700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The regression explains 84% of the variation." xmpDM:startTime="2036700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="At this confidence level, we can accept the hypothesis that the relationship is linear since the confidence interval for b1" xmpDM:startTime="2036700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="includes 1." xmpDM:startTime="2036700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="2036700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2037267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Let us fit the following curvilinear model to this data: Taking a log of both sides we get: TP_tmp" xmpDM:startTime="2037267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="2037267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2037900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Amdahl's law: I/O rate is proportional to the processor speed." xmpDM:startTime="2037900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For each instruction executed there is one bit of I/O on the average." xmpDM:startTime="2037900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="2037900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2038967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Let us fit the following curvilinear model to this data: Taking a log of both sides we get: TP_tmp" xmpDM:startTime="2038967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="2038967" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2039267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Both coefficients are significant at 90% confidence level." xmpDM:startTime="2039267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The regression explains 84% of the variation." xmpDM:startTime="2039267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="At this confidence level, we can accept the hypothesis that the relationship is linear since the confidence interval for b1" xmpDM:startTime="2039267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="includes 1." xmpDM:startTime="2039267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="2039267" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2039633" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Errors in log I/O rate do seem to be normally distributed." xmpDM:startTime="2039633" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TPps2bmp" xmpDM:startTime="2039633" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2045900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Both coefficients are significant at 90% confidence level." xmpDM:startTime="2045900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="The regression explains 84% of the variation." xmpDM:startTime="2045900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="At this confidence level, we can accept the hypothesis that the relationship is linear since the confidence interval for b1" xmpDM:startTime="2045900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="includes 1." xmpDM:startTime="2045900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="2045900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2064433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Cont) Errors in log I/O rate do seem to be normally distributed." xmpDM:startTime="2064433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TPps2bmp" xmpDM:startTime="2064433" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations Transformation: Some function of the measured response variable y is used." xmpDM:startTime="2104200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For example, Transformation is a subset of the curvilinear regression." xmpDM:startTime="2104200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="However, the ideas apply to non-regression model as well." xmpDM:startTime="2104200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Physical considerations Þ Transformation For example, if response = inter-arrival times y and it is known that the number of" xmpDM:startTime="2104200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="requests per unit time (1/y) has a linear relationship to a predictor If the range of the data covers several" xmpDM:startTime="2104200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="orders of magnitude and the sample size is small." xmpDM:startTime="2104200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="That is, if is large." xmpDM:startTime="2104200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If the homogeneous variance (homoscedasticity) assumption of the residuals is violated." xmpDM:startTime="2104200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="2104200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations (Cont) scatter plot shows non-homogeneous spread Þ Residuals are still functions of the predictors Plot the standard deviation of" xmpDM:startTime="2297933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="residuals at each value of as a function of the mean ." xmpDM:startTime="2297933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If s and the mean : Then a transformation of the form: may help solve the problem TP_tmp TP_tmp TP_tmp" xmpDM:startTime="2297933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp" xmpDM:startTime="2297933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations (Cont) scatter plot shows non-homogeneous spread Þ Residuals are still functions of the predictors Plot the standard deviation of" xmpDM:startTime="2323700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="residuals at each value of as a function of the mean ." xmpDM:startTime="2323700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If s and the mean : Then a transformation of the form: may help solve the problem TP_tmp TP_tmp TP_tmp" xmpDM:startTime="2323700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp" xmpDM:startTime="2323700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="4." xmpDM:startTime="2324367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Constant Standard Deviation of Errors Also known as homoscedasticity Trend  Try curvilinear regression or transformation chp_14_10a chp_14_10b ei ei" xmpDM:startTime="2324367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="2324367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="4." xmpDM:startTime="2383833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Constant Standard Deviation of Errors Also known as homoscedasticity Trend  Try curvilinear regression or transformation chp_14_10a chp_14_10b ei ei" xmpDM:startTime="2383833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="2383833" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations Transformation: Some function of the measured response variable y is used." xmpDM:startTime="2383933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For example, Transformation is a subset of the curvilinear regression." xmpDM:startTime="2383933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="However, the ideas apply to non-regression model as well." xmpDM:startTime="2383933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Physical considerations Þ Transformation For example, if response = inter-arrival times y and it is known that the number of" xmpDM:startTime="2383933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="requests per unit time (1/y) has a linear relationship to a predictor If the range of the data covers several" xmpDM:startTime="2383933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="orders of magnitude and the sample size is small." xmpDM:startTime="2383933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="That is, if is large." xmpDM:startTime="2383933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If the homogeneous variance (homoscedasticity) assumption of the residuals is violated." xmpDM:startTime="2383933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="2383933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations (Cont) scatter plot shows non-homogeneous spread Þ Residuals are still functions of the predictors Plot the standard deviation of" xmpDM:startTime="2404533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="residuals at each value of as a function of the mean ." xmpDM:startTime="2404533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If s and the mean : Then a transformation of the form: may help solve the problem TP_tmp TP_tmp TP_tmp" xmpDM:startTime="2404533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp TP_tmp" xmpDM:startTime="2404533" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations Log Transformation: Standard deviation s is a linear function of the mean (s = a ) w =" xmpDM:startTime="2519467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ln y and, therefore: TP_tmp TP_tmp TP_tmp chp_15_02a" xmpDM:startTime="2519467" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations (Cont) Shifting: y+c (with some suitable c) may be used in place of y." xmpDM:startTime="2697233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful if there are negative or zero values and if the transformation function is not defined for these values." xmpDM:startTime="2697233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="2697233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations Log Transformation: Standard deviation s is a linear function of the mean (s = a ) w =" xmpDM:startTime="2698667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ln y and, therefore: TP_tmp TP_tmp TP_tmp chp_15_02a" xmpDM:startTime="2698667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations (Cont) Shifting: y+c (with some suitable c) may be used in place of y." xmpDM:startTime="2699900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful if there are negative or zero values and if the transformation function is not defined for these values." xmpDM:startTime="2699900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="2699900" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations Log Transformation: Standard deviation s is a linear function of the mean (s = a ) w =" xmpDM:startTime="2702300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ln y and, therefore: TP_tmp TP_tmp TP_tmp chp_15_02a" xmpDM:startTime="2702300" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations (Cont) Shifting: y+c (with some suitable c) may be used in place of y." xmpDM:startTime="2728667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful if there are negative or zero values and if the transformation function is not defined for these values." xmpDM:startTime="2728667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="2728667" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Box-Cox Transformations If the value of the exponent a in a power transformation is not known, Box-Cox family of transformations" xmpDM:startTime="2867367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="can be used: Where g is the geometric mean of the responses: The Box-Cox transformation has the property that w" xmpDM:startTime="2867367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="has the same units as the response y for all values of the exponent a." xmpDM:startTime="2867367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Use a that gives the smallest SSE." xmpDM:startTime="2867367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="2867367" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Outliers Any observation that is atypical of the remaining observations may be considered an outlier." xmpDM:startTime="2907767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Including the outlier in the analysis may change the conclusions significantly." xmpDM:startTime="2907767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Excluding the outlier from the analysis may lead to a misleading conclusion, if the outlier in fact represents a correct" xmpDM:startTime="2907767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="observation of the system behavior." xmpDM:startTime="2907767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="A number of statistical tests have been proposed to test if a particular value is an outlier." xmpDM:startTime="2907767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Most of these tests assume a certain distribution for the observations." xmpDM:startTime="2907767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If the observations do not satisfy the assumed distribution, the results of the statistical test would be misleading." xmpDM:startTime="2907767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Easiest way to identify outliers is to look at the scatter plot of the data." xmpDM:startTime="2907767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="2907767" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Box-Cox Transformations If the value of the exponent a in a power transformation is not known, Box-Cox family of transformations" xmpDM:startTime="2909733" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="can be used: Where g is the geometric mean of the responses: The Box-Cox transformation has the property that w" xmpDM:startTime="2909733" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="has the same units as the response y for all values of the exponent a." xmpDM:startTime="2909733" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Use a that gives the smallest SSE." xmpDM:startTime="2909733" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="2909733" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations (Cont) Shifting: y+c (with some suitable c) may be used in place of y." xmpDM:startTime="3043700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful if there are negative or zero values and if the transformation function is not defined for these values." xmpDM:startTime="3043700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp" xmpDM:startTime="3043700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Box-Cox Transformations If the value of the exponent a in a power transformation is not known, Box-Cox family of transformations" xmpDM:startTime="3058800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="can be used: Where g is the geometric mean of the responses: The Box-Cox transformation has the property that w" xmpDM:startTime="3058800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="has the same units as the response y for all values of the exponent a." xmpDM:startTime="3058800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Use a that gives the smallest SSE." xmpDM:startTime="3058800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="TP_tmp TP_tmp" xmpDM:startTime="3058800" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Outliers Any observation that is atypical of the remaining observations may be considered an outlier." xmpDM:startTime="3067600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Including the outlier in the analysis may change the conclusions significantly." xmpDM:startTime="3067600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Excluding the outlier from the analysis may lead to a misleading conclusion, if the outlier in fact represents a correct" xmpDM:startTime="3067600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="observation of the system behavior." xmpDM:startTime="3067600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="A number of statistical tests have been proposed to test if a particular value is an outlier." xmpDM:startTime="3067600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Most of these tests assume a certain distribution for the observations." xmpDM:startTime="3067600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="If the observations do not satisfy the assumed distribution, the results of the statistical test would be misleading." xmpDM:startTime="3067600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Easiest way to identify outliers is to look at the scatter plot of the data." xmpDM:startTime="3067600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="3067600" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Outliers (Cont) Any value significantly away from the remaining observations should be investigated for possible experimental errors." xmpDM:startTime="3232200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Other experiments in the neighborhood of the outlying observation may be conducted to verify that the response is typical of" xmpDM:startTime="3232200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="the system behavior in that operating region." xmpDM:startTime="3232200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Once the possibility of errors in the experiment has been eliminated, the analyst may decide to include or exclude the" xmpDM:startTime="3232200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="suspected outlier based on intuition." xmpDM:startTime="3232200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="One alternative is to repeat the analysis with and without the outlier and state the results separately." xmpDM:startTime="3232200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Another alternative is to divide the operating region into two (or more) sub-regions and obtain a separate model for each" xmpDM:startTime="3232200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="sub-region." xmpDM:startTime="3232200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="3232200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression Not verifying that the relationship is linear." xmpDM:startTime="3400633" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Relying on automated results without visual verification chp_15_05 In all these cases, R2 = High High R2 is necessary but" xmpDM:startTime="3400633" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="not sufficient for a good model." xmpDM:startTime="3400633" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="3400633" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression (Cont) Attaching importance to numerical values of regression parameters." xmpDM:startTime="3429233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="CPU time in seconds = 0.01" xmpDM:startTime="3429233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Number of disk I/O's) + 0.001" xmpDM:startTime="3429233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Memory size in kilobytes) 0.001" xmpDM:startTime="3429233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="is too small &gt; memory size can be ignored CPU time in milliseconds = 10 (Number of disk I/O's) +" xmpDM:startTime="3429233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="1 (Memory size in kilobytes) CPU time in seconds = 0.01" xmpDM:startTime="3429233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Number of disk I/O's) + 1 (Memory size in Mbytes) Not specifying confidence intervals for the regression parameters." xmpDM:startTime="3429233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Not specifying the coefficient of determination." xmpDM:startTime="3429233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="3429233" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression Not verifying that the relationship is linear." xmpDM:startTime="3430933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Relying on automated results without visual verification chp_15_05 In all these cases, R2 = High High R2 is necessary but" xmpDM:startTime="3430933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="not sufficient for a good model." xmpDM:startTime="3430933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="3430933" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression (Cont) Attaching importance to numerical values of regression parameters." xmpDM:startTime="3436700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="CPU time in seconds = 0.01" xmpDM:startTime="3436700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Number of disk I/O's) + 0.001" xmpDM:startTime="3436700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Memory size in kilobytes) 0.001" xmpDM:startTime="3436700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="is too small &gt; memory size can be ignored CPU time in milliseconds = 10 (Number of disk I/O's) +" xmpDM:startTime="3436700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="1 (Memory size in kilobytes) CPU time in seconds = 0.01" xmpDM:startTime="3436700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="(Number of disk I/O's) + 1 (Memory size in Mbytes) Not specifying confidence intervals for the regression parameters." xmpDM:startTime="3436700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Not specifying the coefficient of determination." xmpDM:startTime="3436700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="" xmpDM:startTime="3436700" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression (Cont) Confusing the coefficient of determination and the coefficient of correlation R=Coefficient of correlation, R2= Coefficient" xmpDM:startTime="3476200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="of determination R=0.8," xmpDM:startTime="3476200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="R2=0.64" xmpDM:startTime="3476200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Þ Regression explains only 64% of variation and not 80%." xmpDM:startTime="3476200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Using highly correlated variables as predictor variables." xmpDM:startTime="3476200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Analysts often start a multi-linear regression with as many predictor variables as possible Þ severe multicollinearity problems." xmpDM:startTime="3476200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Using regression to predict far beyond the measured range." xmpDM:startTime="3476200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Predictions should be specified along with their confidence intervals Using too many predictor variables." xmpDM:startTime="3476200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="k predictors Þ 2k-1 subsets" xmpDM:startTime="3476200" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression (Cont) Subset giving the maximum R2 is the best." xmpDM:startTime="3539333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="But, other subsets that are close may be used instead for practical or engineering reasons." xmpDM:startTime="3539333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="For example, if the second best has only one variable compared to five in the best, the second best may" xmpDM:startTime="3539333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="the preferred model." xmpDM:startTime="3539333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="10." xmpDM:startTime="3539333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Measuring only a small subset of the complete range of operation, e.g.," xmpDM:startTime="3539333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="10 or 20 users on a 100 user system." xmpDM:startTime="3539333" xmpDM:duration="1000"/></rdf:li><rdf:li><rdf:Description xmpDM:name="chp_15_06" xmpDM:startTime="3539333" xmpDM:duration="1000"/></rdf:li></rdf:Seq>
                     </xmpDM:markers>
                  </rdf:Description>
               </rdf:li>
               <rdf:li>
                  <rdf:Description xmpDM:trackType="TableOfContents" xmpDM:frameRate="f1000" xmpDM:trackName="Table of Contents">
                     <xmpDM:markers>
                        <rdf:Seq>
                           <rdf:li><rdf:Description xmpDM:name="Problem of Multicollinearity" xmpDM:startTime="0" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="0" tscDM:imageoffset="0" tscDM:imagerect="0, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.2" xmpDM:startTime="58833" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="1" tscDM:imageoffset="0" tscDM:imagerect="75, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Problem of Multicollinearity" xmpDM:startTime="83433" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="2" tscDM:imageoffset="0" tscDM:imagerect="150, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3" xmpDM:startTime="96367" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="3" tscDM:imageoffset="0" tscDM:imagerect="225, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3 (Cont)" xmpDM:startTime="173333" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="4" tscDM:imageoffset="0" tscDM:imagerect="300, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A" xmpDM:startTime="278233" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="5" tscDM:imageoffset="0" tscDM:imagerect="375, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont)" xmpDM:startTime="315600" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="6" tscDM:imageoffset="0" tscDM:imagerect="450, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A" xmpDM:startTime="321467" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="7" tscDM:imageoffset="0" tscDM:imagerect="525, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Presentation Paused" xmpDM:startTime="341233" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="8" tscDM:imageoffset="0" tscDM:imagerect="600, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Presentation Resumed" xmpDM:startTime="341300" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="9" tscDM:imageoffset="0" tscDM:imagerect="675, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont)" xmpDM:startTime="368167" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="10" tscDM:imageoffset="0" tscDM:imagerect="750, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A" xmpDM:startTime="385300" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="11" tscDM:imageoffset="0" tscDM:imagerect="825, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont)" xmpDM:startTime="416667" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="12" tscDM:imageoffset="0" tscDM:imagerect="900, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A" xmpDM:startTime="419967" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="13" tscDM:imageoffset="0" tscDM:imagerect="975, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3 (Cont)" xmpDM:startTime="431100" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="14" tscDM:imageoffset="0" tscDM:imagerect="1050, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3" xmpDM:startTime="431267" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="15" tscDM:imageoffset="0" tscDM:imagerect="1125, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Problem of Multicollinearity" xmpDM:startTime="431400" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="16" tscDM:imageoffset="0" tscDM:imagerect="1200, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.2" xmpDM:startTime="431567" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="17" tscDM:imageoffset="0" tscDM:imagerect="1275, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ANOVA Table  for Multiple Linear Regression" xmpDM:startTime="431733" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="18" tscDM:imageoffset="0" tscDM:imagerect="1350, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name=" F-Test (Cont)" xmpDM:startTime="431900" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="19" tscDM:imageoffset="0" tscDM:imagerect="1425, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="F-Test " xmpDM:startTime="432100" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="20" tscDM:imageoffset="0" tscDM:imagerect="1500, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name=" Analysis of Variance" xmpDM:startTime="432300" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="21" tscDM:imageoffset="0" tscDM:imagerect="1575, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1 (Cont)" xmpDM:startTime="432500" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="22" tscDM:imageoffset="0" tscDM:imagerect="1650, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1 (Cont)" xmpDM:startTime="432700" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="23" tscDM:imageoffset="0" tscDM:imagerect="1725, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1 (Cont)" xmpDM:startTime="433033" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="24" tscDM:imageoffset="0" tscDM:imagerect="1800, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1 (Cont)" xmpDM:startTime="433433" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="25" tscDM:imageoffset="0" tscDM:imagerect="1875, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1 (Cont)" xmpDM:startTime="465533" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="26" tscDM:imageoffset="0" tscDM:imagerect="1950, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1 (Cont)" xmpDM:startTime="465800" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="27" tscDM:imageoffset="0" tscDM:imagerect="2025, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.1 (Cont)" xmpDM:startTime="466033" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="28" tscDM:imageoffset="0" tscDM:imagerect="2100, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name=" Analysis of Variance" xmpDM:startTime="466400" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="29" tscDM:imageoffset="0" tscDM:imagerect="2175, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="F-Test " xmpDM:startTime="466833" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="30" tscDM:imageoffset="0" tscDM:imagerect="2250, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name=" F-Test (Cont)" xmpDM:startTime="467233" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="31" tscDM:imageoffset="0" tscDM:imagerect="2325, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="ANOVA Table  for Multiple Linear Regression" xmpDM:startTime="467633" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="32" tscDM:imageoffset="0" tscDM:imagerect="2400, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.2" xmpDM:startTime="468133" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="33" tscDM:imageoffset="0" tscDM:imagerect="2475, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Problem of Multicollinearity" xmpDM:startTime="468533" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="34" tscDM:imageoffset="0" tscDM:imagerect="2550, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3" xmpDM:startTime="469067" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="35" tscDM:imageoffset="0" tscDM:imagerect="2625, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.3 (Cont)" xmpDM:startTime="469567" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="36" tscDM:imageoffset="0" tscDM:imagerect="2700, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A" xmpDM:startTime="470033" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="37" tscDM:imageoffset="0" tscDM:imagerect="2775, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont)" xmpDM:startTime="471167" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="38" tscDM:imageoffset="0" tscDM:imagerect="2850, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A" xmpDM:startTime="513100" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="39" tscDM:imageoffset="0" tscDM:imagerect="2925, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Homework 15A (Cont)" xmpDM:startTime="518233" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="40" tscDM:imageoffset="0" tscDM:imagerect="3000, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Regression with Categorical Predictors " xmpDM:startTime="572800" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="41" tscDM:imageoffset="0" tscDM:imagerect="3075, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="727600" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="42" tscDM:imageoffset="0" tscDM:imagerect="3150, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="777967" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="43" tscDM:imageoffset="0" tscDM:imagerect="3225, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="810933" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="44" tscDM:imageoffset="0" tscDM:imagerect="3300, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="817867" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="45" tscDM:imageoffset="0" tscDM:imagerect="3375, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="825700" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="46" tscDM:imageoffset="0" tscDM:imagerect="3450, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="837433" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="47" tscDM:imageoffset="0" tscDM:imagerect="3525, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="856533" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="48" tscDM:imageoffset="0" tscDM:imagerect="3600, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="858000" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="49" tscDM:imageoffset="0" tscDM:imagerect="3675, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="874933" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="50" tscDM:imageoffset="0" tscDM:imagerect="3750, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="924767" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="51" tscDM:imageoffset="0" tscDM:imagerect="3825, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="928167" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="52" tscDM:imageoffset="0" tscDM:imagerect="3900, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="945167" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="53" tscDM:imageoffset="0" tscDM:imagerect="3975, 0, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="948700" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="54" tscDM:imageoffset="0" tscDM:imagerect="50, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="950667" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="55" tscDM:imageoffset="0" tscDM:imagerect="125, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="1038967" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="56" tscDM:imageoffset="0" tscDM:imagerect="200, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="1069467" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="57" tscDM:imageoffset="0" tscDM:imagerect="275, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name=" Case Study 15.1: RPC performance" xmpDM:startTime="1070100" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="58" tscDM:imageoffset="0" tscDM:imagerect="350, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1 (Cont)" xmpDM:startTime="1070533" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="59" tscDM:imageoffset="0" tscDM:imagerect="425, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name=" Case Study 15.1: RPC performance" xmpDM:startTime="1071367" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="60" tscDM:imageoffset="0" tscDM:imagerect="500, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="         Categorical Predictors (Cont)" xmpDM:startTime="1072667" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="61" tscDM:imageoffset="0" tscDM:imagerect="575, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name=" Case Study 15.1: RPC performance" xmpDM:startTime="1074233" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="62" tscDM:imageoffset="0" tscDM:imagerect="650, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1 (Cont)" xmpDM:startTime="1145867" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="63" tscDM:imageoffset="0" tscDM:imagerect="725, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name=" Case Study 15.1: RPC performance" xmpDM:startTime="1270533" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="64" tscDM:imageoffset="0" tscDM:imagerect="800, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1 (Cont)" xmpDM:startTime="1290533" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="65" tscDM:imageoffset="0" tscDM:imagerect="875, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name=" Case Study 15.1: RPC performance" xmpDM:startTime="1310033" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="66" tscDM:imageoffset="0" tscDM:imagerect="950, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Case Study 15.1 (Cont)" xmpDM:startTime="1347100" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="67" tscDM:imageoffset="0" tscDM:imagerect="1025, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Differing Conclusions" xmpDM:startTime="1350467" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="68" tscDM:imageoffset="0" tscDM:imagerect="1100, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="          Differing Conclusions (Cont)" xmpDM:startTime="1502433" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="69" tscDM:imageoffset="0" tscDM:imagerect="1175, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Curvilinear Regression" xmpDM:startTime="1540400" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="70" tscDM:imageoffset="0" tscDM:imagerect="1250, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="  Curvilinear Regression: Other Examples" xmpDM:startTime="1644300" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="71" tscDM:imageoffset="0" tscDM:imagerect="1325, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="1795333" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="72" tscDM:imageoffset="0" tscDM:imagerect="1400, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="1883233" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="73" tscDM:imageoffset="0" tscDM:imagerect="1475, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="1930967" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="74" tscDM:imageoffset="0" tscDM:imagerect="1550, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="2012700" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="75" tscDM:imageoffset="0" tscDM:imagerect="1625, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="2036700" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="76" tscDM:imageoffset="0" tscDM:imagerect="1700, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="2037267" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="77" tscDM:imageoffset="0" tscDM:imagerect="1775, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4" xmpDM:startTime="2037900" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="78" tscDM:imageoffset="0" tscDM:imagerect="1850, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="2038967" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="79" tscDM:imageoffset="0" tscDM:imagerect="1925, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="2039267" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="80" tscDM:imageoffset="0" tscDM:imagerect="2000, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="2039633" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="81" tscDM:imageoffset="0" tscDM:imagerect="2075, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="2045900" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="82" tscDM:imageoffset="0" tscDM:imagerect="2150, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Example 15.4 (Cont)" xmpDM:startTime="2064433" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="83" tscDM:imageoffset="0" tscDM:imagerect="2225, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations" xmpDM:startTime="2104200" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="84" tscDM:imageoffset="0" tscDM:imagerect="2300, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations (Cont)" xmpDM:startTime="2297933" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="85" tscDM:imageoffset="0" tscDM:imagerect="2375, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="End of presentation" xmpDM:startTime="2299333" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="86" tscDM:imageoffset="0" tscDM:imagerect="2450, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations (Cont)" xmpDM:startTime="2323700" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="87" tscDM:imageoffset="0" tscDM:imagerect="2525, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="4. Constant Standard Deviation of Errors" xmpDM:startTime="2324367" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="88" tscDM:imageoffset="0" tscDM:imagerect="2600, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="End of presentation" xmpDM:startTime="2367733" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="89" tscDM:imageoffset="0" tscDM:imagerect="2675, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="4. Constant Standard Deviation of Errors" xmpDM:startTime="2383833" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="90" tscDM:imageoffset="0" tscDM:imagerect="2750, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations" xmpDM:startTime="2383933" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="91" tscDM:imageoffset="0" tscDM:imagerect="2825, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Transformations (Cont)" xmpDM:startTime="2404533" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="92" tscDM:imageoffset="0" tscDM:imagerect="2900, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations" xmpDM:startTime="2519467" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="93" tscDM:imageoffset="0" tscDM:imagerect="2975, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="       Useful Transformations (Cont)" xmpDM:startTime="2697233" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="94" tscDM:imageoffset="0" tscDM:imagerect="3050, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations" xmpDM:startTime="2698667" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="95" tscDM:imageoffset="0" tscDM:imagerect="3125, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="       Useful Transformations (Cont)" xmpDM:startTime="2699900" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="96" tscDM:imageoffset="0" tscDM:imagerect="3200, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Useful Transformations" xmpDM:startTime="2702300" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="97" tscDM:imageoffset="0" tscDM:imagerect="3275, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="       Useful Transformations (Cont)" xmpDM:startTime="2728667" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="98" tscDM:imageoffset="0" tscDM:imagerect="3350, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Box-Cox Transformations" xmpDM:startTime="2867367" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="99" tscDM:imageoffset="0" tscDM:imagerect="3425, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Outliers" xmpDM:startTime="2907767" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="100" tscDM:imageoffset="0" tscDM:imagerect="3500, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Box-Cox Transformations" xmpDM:startTime="2909733" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="101" tscDM:imageoffset="0" tscDM:imagerect="3575, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="       Useful Transformations (Cont)" xmpDM:startTime="3043700" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="102" tscDM:imageoffset="0" tscDM:imagerect="3650, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Box-Cox Transformations" xmpDM:startTime="3058800" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="103" tscDM:imageoffset="0" tscDM:imagerect="3725, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Outliers" xmpDM:startTime="3067600" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="104" tscDM:imageoffset="0" tscDM:imagerect="3800, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Outliers (Cont)" xmpDM:startTime="3232200" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="105" tscDM:imageoffset="0" tscDM:imagerect="3875, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="    Common Mistakes in Regression" xmpDM:startTime="3400633" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="106" tscDM:imageoffset="0" tscDM:imagerect="3950, 56, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression (Cont)" xmpDM:startTime="3429233" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="107" tscDM:imageoffset="0" tscDM:imagerect="25, 112, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="    Common Mistakes in Regression" xmpDM:startTime="3430933" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="108" tscDM:imageoffset="0" tscDM:imagerect="100, 112, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression (Cont)" xmpDM:startTime="3436700" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="109" tscDM:imageoffset="0" tscDM:imagerect="175, 112, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression (Cont)" xmpDM:startTime="3476200" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="110" tscDM:imageoffset="0" tscDM:imagerect="250, 112, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Marker 1" xmpDM:startTime="3513667" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="111" tscDM:imageoffset="0" tscDM:imagerect="325, 112, 75, 56"/></rdf:li><rdf:li><rdf:Description xmpDM:name="Common Mistakes in Regression (Cont)" xmpDM:startTime="3539333" tscDM:image="k_15orm2_Thumbnails.png" tscDM:imageindex="112" tscDM:imageoffset="0" tscDM:imagerect="400, 112, 75, 56"/></rdf:li></rdf:Seq>
                     </xmpDM:markers>
                  </rdf:Description>
               </rdf:li>
            </rdf:Bag>
         </xmpDM:Tracks>
         <tscDM:controller>
            <rdf:Description xmpDM:name="tscplayer">
               <tscDM:parameters>
                  <rdf:Bag>
                     <rdf:li xmpDM:name="autohide" xmpDM:value="true"/><rdf:li xmpDM:name="autoplay" xmpDM:value="false"/><rdf:li xmpDM:name="loop" xmpDM:value="false"/><rdf:li xmpDM:name="searchable" xmpDM:value="true"/><rdf:li xmpDM:name="captionsenabled" xmpDM:value="false"/><rdf:li xmpDM:name="sidebarenabled" xmpDM:value="false"/><rdf:li xmpDM:name="unicodeenabled" xmpDM:value="false"/><rdf:li xmpDM:name="backgroundcolor" xmpDM:value="000000"/><rdf:li xmpDM:name="sidebarlocation" xmpDM:value="left"/><rdf:li xmpDM:name="endaction" xmpDM:value="stop"/><rdf:li xmpDM:name="endactionparam" xmpDM:value="true"/><rdf:li xmpDM:name="locale" xmpDM:value="en-US"/></rdf:Bag>
               </tscDM:parameters>
               <tscDM:controllerText>
                  <rdf:Bag>
                  </rdf:Bag>
               </tscDM:controllerText>
            </rdf:Description>
         </tscDM:controller>
         <tscDM:contentList>
            <rdf:Description>
               <tscDM:files>
                  <rdf:Seq>
                     <rdf:li xmpDM:name="0" xmpDM:value="k_15orm2.mp4"/><rdf:li xmpDM:name="1" xmpDM:value="k_15orm2_First_Frame.png"/><rdf:li xmpDM:name="2" xmpDM:value="k_15orm2_Thumbnails.png"/></rdf:Seq>
               </tscDM:files>
            </rdf:Description>
         </tscDM:contentList>
      </rdf:Description>
   </rdf:RDF>
</x:xmpmeta>