In this assignment you will build a logistic regression model to predict alcohol involvement of the driver in the crash (yes/no). You will create a nominal variable that is 1 if the driver has a BAC>=0.08 and 0 otherwise, and you will divide the data set into 65% training, 20% validation and 15% Test.
Review Assignment 4 Report Example .
Note: You will want to create your assignments with a similar format, layout and content. The answers shown are NOT THE REAL ONES and that copying the example will result in a 0
Review the Grading Table (below).
Use the Confusion Matrix spreadsheet for the validation set in the appendix. You can cut and past the information from JMP and then copy the confusion matrix into the appendix of the report.
Refer to the Alcohol Prediction Data jmp file to obtain the data needed for this assignment.
Interpret sensitivity, specificity, false positive and false negative rate, overall error rate.
Add the ROC and Lift curve in the appendix and interpret them in your report.
Provide strengths and weaknesses of the model for prediction as well as which predictors are most important contributors to explaining BAC>=0.08.
Use 10 pt Font Calibri or Times New Roman.
The report must include the following elements:
Figure 1: Receiver Operating Characteristic on Validation Data
Figure 2: Lift Curve on Validation Data
Table 1: Confusion Matrix for Validation Data
Table 2: Odds Ratios (Day_Week, HitRun Description, and Gender)
Table 3: Effect Summary
Include a title for your report (e.g, “Real estate…) and create headings for each section.
