QUESTION 1 Assume that you are interested in modeling the sa

QUESTION 1

Assume that you are interested in modeling the sales for a local Target each month for the last 10 years.   One thing the Target executives are specifically interested in is how sales change from one month to the next.   If you were to code the categorical variable of month into your model how many dummy variables would you need? (record your answer as a number).

QUESTION 2

Assume that you are interested in modeling the sales for a local Target each month for the last 10 years.   One thing the Target executives are specifically interested in is how sales change from one month to the next. In the previous question you determined how many dummy variables you would need? Further assume that month is the only variable you will include in this model.

Although not specifically mentioned in class: The error degrees of freedom for a model with dummy variables is calculated in exactly the same way as another multiple regression model. Df = n - p - 1 where n = number of observations and p - number of coefficients in the model.

If we fit a model with 10 years of data (one record for each month during each of these 10 years) and the only explanatory variable in our model is based on the categorical variable of month: What are the error degrees of freedom for this model?

Solution

QUESTION 1

Assume that you are interested in modeling the sales for a local Target each month for the last 10 years.   One thing the Target executives are specifically interested in is how sales change from one month to the next.   If you were to code the categorical variable of month into your model how many dummy variables would you need? (record your answer as a number).

There are 12 months

dummy variables we need   12-1 = 11

QUESTION 2

Assume that you are interested in modeling the sales for a local Target each month for the last 10 years.   One thing the Target executives are specifically interested in is how sales change from one month to the next. In the previous question you determined how many dummy variables you would need? Further assume that month is the only variable you will include in this model.

Although not specifically mentioned in class: The error degrees of freedom for a model with dummy variables is calculated in exactly the same way as another multiple regression model. Df = n - p - 1 where n = number of observations and p - number of coefficients in the model.

If we fit a model with 10 years of data (one record for each month during each of these 10 years) and the only explanatory variable in our model is based on the categorical variable of month: What are the error degrees of freedom for this model?

Number of observations =n=10*12=120

p=number of variables=11

error degrees of freedom for this model =120-11-1= 108