This problem is based on an exercise from Abraham and Ledolter (2006, pp XXXXXXXXXXwhich focuses on monthly sales from a bookstore in the city of Vienna, Austria. The available data consisted of 93...

This problem is based on an exercise from Abraham and Ledolter (2006, pp. 335-337) which focuses on monthly sales from a bookstore in the city of Vienna, Austria. The available data consisted of 93 consecutive monthly observations on the following variables: Sales = Sales (in hundreds of dollars) Advert = Advertising spend in the current month LaglAdvert = Advertising spend in the previous month Time = Time in months Month_i = Dummy variable which is 1 for month i and 0 otherwise (i = 2,3, ..., 12)
Exercises 329
The data can be found on the book website in the file bookstore.txt.
(a) Follow the advice of Abraham and Ledolter (2006, pp. 336-337) and first build a model for Sales ignoring the effects due to Advert and Lag lAdvert. Ensure that you produce diagnostic plots to justify your choice of model. Describe any weaknesses in your model. (b) Add the effects due to Advert and LaglAdvert to the model you have developed in (a). Last month's advertising (LaglAdvert) is thought to have an impact on the current month's sales. Obtain a final model for predicting Sales. Ensure that you produce diagnostic plots to justify your choice of model. Describe any weaknesses in your model.