3. Continuing with the "infringing sales" data, a second regression was run using log (sales) as the dependent variable instead of sales. The estimates from this regression were as follows: Parameter Estimates Intercept Time Season Coefficient Estimate 0.59015 0.01614 0.05118 Standard Error 0.03351 0.00310 0.01608 The Explained sum of squares is 0.18826, and the total sum of squares is 0.25741. Based on these results, calculate the predicted log sales for each of the bimonthly periods of 1981, and then use either the residual sum of squares or the average of exponentiated residuals to calculate the predicted sales for the bimonthly periods. The procedure is described on pp. 206- 207 of the Wooldridge textbook, at 6.40 and 6.43, where some alternative ways to do it are described. Please indicate the alternative you are using. Predicted Jan/Feb. Mar/Apr May/Jun Jul/Aug Sep/Oct Nov/Dec Actual 2.15 2.40 2.947 3.326 2.789 2.557 3. Continuing with the "infringing sales" data, a second regression was run using log (sales) as the dependent variable instead of sales. The estimates from this regression were as follows: Parameter Estimates Intercept Time Season Coefficient Estimate 0.59015 0.01614 0.05118 Standard Error 0.03351 0.00310 0.01608 The Explained sum of squares is 0.18826, and the total sum of squares is 0.25741. Based on these results, calculate the predicted log sales for each of the bimonthly periods of 1981, and then use either the residual sum of squares or the average of exponentiated residuals to calculate the predicted sales for the bimonthly periods. The procedure is described on pp. 206- 207 of the Wooldridge textbook, at 6.40 and 6.43, where some alternative ways to do it are described. Please indicate the alternative you are using. Predicted Jan/Feb. Mar/Apr May/Jun Jul/Aug Sep/Oct Nov/Dec Actual 2.15 2.40 2.947 3.326 2.789 2.557