Our hedge fund is considering investing $40Million/yr in loans on LendingClub.com. These loans are an appealing investment with rates averaging 17%. However, 5% of invested money on the site is lost due to borrower defaults. You would like to be able to screen out the riskiest borrowers. If successful, your organization will use your approach to justify investing on the site. They will then apply the method in an automated fashion to choose which loans to fund. You begin the project with the data set LendingClub(clean).zip. Your target is whether the loan was “bad” or not.

Administrative

• You should follow all directions in this document
• The write up should be written from the perspective of the group
• The write up should be free from grammatical / spelling errors
Original Model
• Explain the file that you loaded into Data Robot.
• What is the name of the column for your target? What qualifies as a “bad” loan? What type of target are you working with?
• Did you remove any columns because of target leakage?

Model Improvement

• What feature engineering techniques, feature selection, or changes in Data Robot did you try? Be specific about what you did (specific features/ techniques where applicable).
• Why did you think this would create improvement? Did it? How do you know either way? Please provide screen shots to supplement your response.

Feature Analysis

• Overall, what are the top three most “important” features? Please provide a screen shot to supplement your response.
• What features were selected in Data Robot for your best model?

Model Selection

• What was your most accurate model (after improvements)? Please provide a screen shot to supplement your response.
• Was this the “fastest model”? Either way, how fast was it? Please provide a screen shot to supplement your response.
• Do you think this model would improve with more data? Why or why not? Please supplement your response with a screen shot.

Order your paper today and have the best from our writers…