[4/10, 9:01 AM] Qevoh: An analyst has de-seasonalized quarterly sales data for the past five years. S/he applied linear regression to these data, using Sales as a dependent variable and Time as Independent variable. For each quarter in the analysis, Actual sales were subtracted from sales "estimated" by the regression line. The above procedure describes the computation of:

Nave forecasts

Cyclical Residuals

Seasonally adjusted moving average

R2 (coefficient of determination)

Machalonobis D2

[4/10, 9:02 AM] Qevoh: 3. Fit both models obtained in part 2. Obtain the summary given by R for each model and comment.

For the smallest model found in part 2 answer the following question:

4.Find the influential points (using the R command influence.measures(model))

[PART 2 was --> 2. Starting with the regression of HT18 on all six regressors, use backward elimination to find the "best" model using the AIC criteria. (R code: step(model)). Then find the "best" model using the BIC criteria.)

3.8 Exercises 47 Age Weight Pulse Blood pressure 21 71 88 160 24 56 70 135 25 65 72 140 28 53 60 1 18 32 67 134 35 80 70 168 37 57 68 111 38 58 64 124 39 68 75 144 41 60 67 112 41 64 81 138 42 68 78 128 43 54 60 118 45 61 56 134 47 55 64 1 16 54 87 92 152 62 52 74 123 68 85 88 161 73 49 70 112 79 56 69 129 Table 3.1 Blood pressure data.(i) One R2 is 0.03% and the other is 0.07%. Which R2 goes with the LOW/TIME model (the top graph)? [ Select ] (ii) The R2 for the HIGH/LOW model is low (regardless of whether it is 0.03% or 0.07%), so the best prediction for HIGH when LOW is 17 meters would be [ Select ] the input of the HIGH/LOW model when 17 is the output the output of the HIGH/LOW model when 17 is the input the average HIGH value the average LOW value