Overview

What variables were significant?

We found the day of the week to be a significant factor on Daily Sales. The F subset test results:
f_calc = 39.41 and f_crit = 2.76, therefore we reject the null that the betas are not significant.  We can conclude that at least one of the betas are strongly significant towards sales.

We ran a t test to test if snowfall was significant. 
t_calc = -0.41 and t_crit = -1.671.  We could not reject the null meaning snowfall is not significant towards sales.

We found that the month that sales are generated in is a significant factor.  t_calc = 1.94 and t_crit = 1.671.  The test statistic falls in the rejection region meaning it has a significant effect.  

What problems did the group encounter?

We ran into the problem of not using data that covered the time span equivalent to one year. Because January is the slack month and also has coupons available, it was being compared to months with higher sales. This caused our coupon coefficient to be negative instead of positve.

Another problem we encountered was Perfect Multicollinearity. Some of our previous explanatory variables were changing at the same time. For example, we had a dummy variable for the month of January (1 if January, 0 if otherwise). But we deleted that variable from our study because it was changing with the dummy variable for coupons, causing perfect multicollinearity.

Also, we couldn't decide what the competitors should be limited to (i.e. only Chinese restaurants, other sit-down restaurants, fast-food, etc). So we decided to exclude competitors as an explanatory variable.

Determining how much snowfall should be present for us to input it into the program was difficult to decide.  Therefore, we decided to assign snowfall if there was any present at all.

These problems biased some values in our results, such as the coupons. Snowfall would have impacted sales more if we only included heavy snowfall, freezing rain, etc.

What were the group's predictions toward the significant variables?

• We predicted Friday Saturday would generate more Sales. 



• If snowfall was present, Sales would decrease due to the bad weather.
 


• If coupons were available, sales would increase.

If there was more time, what else would have been included in this study?

We would have included at least a year's worth of data so that we could tell how much coupons actually increase Sales. Also, if the neighborhood wealth has an effect on sales.

Also, we could have investigated the competitor's variable to have the capability of adding this into the study.