Question

II 1. The Advertising data set consists of the sales (in thousands of units) of a particular product in 400 different markets. It also contains the advertising budgets (in thousands of dollars) for the product in each of the markets for three different media: TV, radio, and newspaper. The data set is divided in two parts-a training set consisting of 200 observations and a test set consisting of the remaining 200 observations. Three models are used on training data and the results are provided in the following table along with test error. Table 1: Summary of Regression Analysis Model 1 Model 2 Model 3 4.6 Intercept 10 12.7 0.2 Price -0.7 2.3 Media[TV] 2.9 1.6 Media[Radio) 1.9 0.68 056 0.71 R-sg 0.67 Adj. Rosa 0.56 0.71 1.7 1.5 0.6 Test error Answer below questions now: 1. The fitted regression equation for Model 1 2. interpretation of R-sa 3. predict sales when price 2.3 and the media used is TV 4. which of the models is better in terms of test error | ENG 10:06 01-08-20 o w

Answer 1

(1) For model 1, the regression line is: $\widehat{Sales}$ = 4.6 + (- 0.2) .

(2) R-sq value tells us the percentage of variation in the dependent variable (Sales) that can be explained by its regression on the independent variables (Price, Media (TV) and Media (Radio)).

(3) Here, Price = 2.3 and Media (TV) = 1. We should use Model 3 for this purpose, since this model contains both of these independent variables. Hence, predicted sales = 12.7 + (-0.7 * 2.3) + (2.9 * 1) = 13.99 thousand units.

(4) Model 3 is the best, since it has the least test error.

Price