Please allocate sufficient time for all homework assignments as they usually take time to complete. Start early on assignments. If you have doubts, please email the instructor.

Please submit your answers and required snapshots of reports and charts as MS Word (or PDF) file, together with your Excel workbook.

Use the file of summer products with rating and performance to build a model and find factors that could affect whether the product is popular.

Variable definitions could be found in the file.

Run the logistic regression, don't forget to split data into training and validation datasets, and use stepwise for variable selection. Choose the best model among models recommended by variable selection. Report the selected model's

1. Regression Model (write down the model in three forms and paste the coefficients table)

1. Explain the coefficients (all of them)

1. How does this model work? (hint: check the ROC curve and compare ROC curve of training and validation)

1. Using validation score, what's the error rate of class 0, the error rate of class 1, the precision, and recall. Please also show the calculation steps.

1. Do the results help solve the problem to identify which products are more likely to be popular? If not, why?

1. Using the validation decile chart to choose a new cutoff value. (hint: consecutively higher than 1). Run the regression again using the new cutoff value. Using validation score, report the new error rate of class 0, error rate of class 1, precision, and recall. Please also show the calculation steps.

1. Using the results of question 6, explain precision and recall in your own words.