首页 > > 详细

Assignment Case: German credit

 Assignment

Case: German credit
The German Credit data set contains observations on 30 variables for 1000 past applicants for 
credit. Each applicant was rated as “good credit” (700 cases) or “bad credit” (300 cases). 
New applicants for credit can also be evaluated on these 30 "predictor" variables. We want to 
develop a credit scoring rule that can be used to determine if a new applicant is a good credit risk 
or a bad credit risk, based on values for one or more of the predictor variables. The data has been 
organized in the spreadsheet GermanCredit.xlsx. All the variables are explained in ‘Codelist’ 
worksheet of the data file. 
The consequences of misclassification have been assessed as follows: the costs of a false 
negative (incorrectly saying an applicant is a good credit risk) outweigh the cost of a false 
positive (incorrectly saying an applicant is a bad credit risk) by a factor of five. This can be 
summarized in the following Table 1. 
Table 1 Opportunity Cost 
Predicted (Decision)
Actual Good (Accept) Bad (Reject)
Good 0 $100
Bad $500 0
The opportunity cost table was derived from the average net profit per loan as shown below:
Table 2 Average Net Profit
Predicted (Decision)
Actual Good (Accept) Bad (Reject)
Good $100 0
Bad $-500 0
Let us use this table in assessing the performance of a logistic regression model because it is 
simpler to explain to decision-makers who are used to thinking of their decision in terms of net 
profits.
1. Review the predictor variables and guess from their definition at what their role might be in a
credit decision. Are there any surprises in the data?
2. Divide the data randomly into training (60%) and test (40%) partitions, and develop a 
classification model using the logistic regression technique in Python and evaluate the model by 
using the confusion matrix and the ROC curve.
3. Based on the confusion matrix and the payoff matrix, what is the net profit on the test data?
4. Let's see if we can improve our performance by changing the cutoff. Rather than accepting the 
above classification of everyone's credit status, let's use the "predicted probability of finding a 
good applicant" in logistic regression as a basis for selecting the best credit risks first, followed 
by poorer risk applicants. 
a. Sort the validation data on "predicted probability of finding a good applicant." 
b. For each test case, calculate the actual cost/gain of extending credit. 
c. Add another column for cumulative net profit. 
d. How far into the test data do you go to get maximum net profit? (Often this is 
specified as a percentile or rounded to deciles.) 
e. If this logistic regression model is scored to future applicants, what "probability of 
success" cutoff should be used in extending credit?
Submission Guidelines
Please submit your work via Blackboard by the deadline. When you submit via Blackboard, you 
should submit your Jupyter notebook with codes and comments.
联系我们 - QQ: 99515681 微信:codinghelp
程序辅导网!