Experian Project

1
Experian Project
Edward Wong University of California, Irvine
2
Overview

• What are we trying to do?
• Predict small business failures
• Why is this important?
• Small Businesses Generates 50 of the private,
  nonfarm GDP in the USA
• (Source Office of Advocacy of the U.S. Small
  Business Administration)

3
My Role The Statistics Guy

• What did I do?
• Why is this important?

4
Problems Encountered

• Missing Data
• Data Points that did not make sense

5
Interesting Find
Variable CTP039 - "Pct of Total Combined Balance
to Combined Trades' total Highest Balance in past
12 months In other words, percentage of credit
used
6
Q1 Successful Business
7
(No Transcript)
8
Interactions with theNaïve Bayes Classifier

• Uses discrete variables
• My Job Manually classify the variables

9

• A 0
• B 1
• C 2
• D 3 or higher

10
(No Transcript)
11
Interactions with theNaïve Bayes Classifier

• Taking it a step further...
• Combining Variables to consider time lapse
  information

12
(No Transcript)