Based on the data we collected, our expected regression equation is:
Walmart revenue = β0 – β1CPI + β2CSI + β3ECOM – β4GAS + β5POP –β6INT – β7UNEMP + β8EARN
Then we ran the regression on Minitab and got an actual regression equation. We used the method (i.e. five questions and nine problems) to evaluate the equation Minitab produced. For each of the nine problems, we ask five questions: What is the problem? How can I detect it? What are the consequences of the problem? What are the possible fixes to the problem? What are the consequences of the fixes?
Problem 1: The residuals should be normally distributed. This can be detected visually using the standardized normal probability plot and histogram. If all the residuals are not within two standard deviations in the normal probability plot or form a bell curve in the histogram, then they are not normally distributed. When the residuals are not normally distributed, it means the …show more content…
Then run a regression model using the residuals square as the dependent variable and fits square as independent variable. The regression will produce a p-value for the fits square, which will be compared to the selected alpha level (it could be 1%, 5%, or 10%). If the p-value is less than the alpha level, we reject the null hypothesis in favor of the alternate. Therefore, the residuals are heteroscedastic. However, if the p-value is greater than the alpha level, we accept the null hypothesis that the residuals are homoscedastic. Heteroscedasticity increases the standard error coefficient just like serial correlation. To resolve the problem of heteroscedasticity, we collect more data or develop a better model to explain the changes in the dependent variable. It is important to note that the KB test is not the most powerful test for heteroscedasticity. The White General test is more powerful in testing for heteroscedasticity. Unfortunately, it is not available on