Module 3, Homework 2

This data file contains 10 predictor variables and a binary response. We are interested in the binary response = 1, and we wish to build a model that has the highest response percentage for the highest decile of predicted probabilities. Try several models in Enterprise Miner, select your best model, and apply it to this small data set using Scoring code. Use the Reporter Node to generate a report of your process.

Hand in a written summary of your modeling process, and print out a few selected pages from your Reporter Node document, such as the workspace diagram, and assessment graphs from your selected model. Also hand in a printout of the predicted probabilities from your best model for the small data set.

To read in either data set, use SAS code like the following:

data probmodel ; 
  infile "h:\stat404\hw2model.txt" ;
  input  x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 bad   ;
  run ;

Then start Enterprise Miner and create an Input Data Node, which in the example above would read from WORK.PROBMODEL. The variables in the small data set are in the same order, but note that the variable 'BAD' is not present in that data set.

one solution to Hw2 (some of the models fitted use knowledge of the variables used to create the data set, so the best solution here is probably better than your solution)