4M Build a regression for the 891 cases that have had a bankruptcy in the past. The modeling of nonbankrupt clients suggests an initial regression model, but you should suspect that other explanatory...

4M

Build a regression for the 891 cases that have had a bankruptcy in the past. The modeling of nonbankrupt clients suggests an initial regression model, but you should suspect that other explanatory variables may be relevant and that coefficients may be different.

(a) Motivation. Explain the purpose of the segmentation approach that models these businesses with a history of bankruptcy from the other businesses.

(b) Method. Which variables that were not available for modeling a business without a bankruptcy become relevant now?

(c) Mechanics. Following the approach used for businesses without a bankruptcy, develop a regression model for predicting fees for those businesses with a bankruptcy. Explain why you did or did not use a test/training sample to validate the modeling.

(d) Message. Compare your model with a comparable model for the nonbankrupt businesses. Use examples to illustrate the differences.