1. Find the appropriate univariate model for the sale of Beer, Wine, and Liquor data (BWL Sales) and justify your answer using data from Jan-2003 to Dec-2006

Calculate the MAPE and RMSE for the data used to run the model and for the hold-up period (7 months)                                                                                                                                                                                                                                              

2. Run a bivariate model using BWL Sales and DPI using data from Jan-2003 to Dec-2006

Compare the results to the univariate model in terms of RMSE and/or MAPE for the data used to run the model and for the hold-up period

3. Account for the trend and seasonality of BWL sales by adding a time index variable and dummies for months 2-12. Run this new model using data from Jan-2003 to Dec-2006

Compare the results to the univariate model in terms of RMSE and/or MAPE for the data used to run the model and for the hold-up period

1. QTR: Quarter (Quarters 1-29 = Control, Quarters 30-50 = Experimental)

2. ACC: Injuries and fatalities from Wednesday to Saturday nighttime accidents

3.FUEL: Fuel consumption (million gallons) in Albuquerque

The Police Department in Alburquerque, New Mexico introduced a van that housed a Blood Alcohol Testing (BAT) device to try to reduce DWI related accidents.

This BATmobile was introduced in Quarter 30 of your data, so you have 29 observations before the program (Control) and 21 during the program (Experimental).

Part of your job is to decide whether or not the program was effective in reducing DWI related accidents.

1. Using the number of accidents (ACC) and the fuel consumption (FUEL), calculate the average number of injuries before and after the program.

Does it look like the program was effective? Explain.

2. Run a multiple-regression model using ACC, FUEL, Quarter dummies (Q2, Q3, Q4), and a dummy variable for whether or not the program was in effect (BAT)

Describe the results for the FUEL and quarter dummies and comment on the evidence related to the efficacy of the BATmobile program.

3. Programs like the BATmobile usually take time to catch on (ramping up). To account for this, modify the BAT dummy variable so that the zero values remain unchanged but the 1's are modified so that the new values are 1, 2, 3, 4, etc. Having this new variable, run the multiple-regression model and comment on whether or not there is evidence of a ramping up effect of the BATmobile program in reducing DWI related accidents.