Question

Answer the following two questions related to regression modeling. a) You are the senior product manager at AMCE Direct. Your analyst has built a regression model predicting a customer’s likelihood of ordering a product. You notice that the model contains two highly correlated variables which are causing multicollinearity problems. In questioning your analyst about this fact, he tells you the only way to rid the equation of this problem is to delete one of the two correlated variables. Do you agree with your analysts comments? Explain your answer fully.

Answer 1

Yes, we do agree with the analyst’s comments because this is a logistic regression problem. The analyst used logistic regression to predict the likelihood of ordering a product but two independent variables in the model are correlated so we can remove one variable from the model to get rid of this problem because removal of one variable would not drastically affect the model accuracy, it would be better to check the accuracy of the model after removing one variable then doing the same with the other variable then select the one which gives better accuracy.