Lending Club Case Study

The goal of this case study is to help finance company in decision making for loan approval based on the applicant’s profile

Table of Contents

• Business Requirements
• Approach
• Technologies Used
• Python Libraries Used
• Conclusions

Business Requirements

• Identify patterns using past loan application data that can trigger charge-off loans
• Risks involved
  • Not approving loans for the applicants who are likely to repay the loan
  • Approving loans for the applicants who are likely to default
• A dataset of loan applications from 2007 to 2011 has to be referred as the base for this analysis

Approach

• Data fields understanding
• Data Pre-processing
• Exploratory Data Analysis

Technologies Used

• Python 3.8
• Jupyter Notebook

Python Libraries Used

• Numpy
• Pandas
• Matplotlib
• Seaborn
• Plotly, Express, Graph Objects, Subplot

Conclusions

• Applicants who have taken the loan for “small business” are having highest probability (27.79%) to default.
• Defaulter Probability is the highest (22.15%) for the applicants having no employment history (Marked as “Unknown” in the graph)
• Probability to default is increasing from grade A to G. Sub-grade F5 has maximum probability (52.5%) to default.
• Low annual income ($25,000 or less) applicants are highest probable (19.59%) to default.
• Probability to default is increasing with increase in Loan Amount. Applicants with Loan amount of $25,000 or above are highly probable to default.
• Probability to default is increasing with increasing Interest Rate and DTI. Loan applicants with interest rate or DTI 19% or above are highly probable to default.
• Applicants from Nebraska (NE) have the highest probability (60%) to default.