Lending Club Case Study

(Assigment by Upgrad and IIIT-B)

The case study focuses on EDA mainly, to understand which major parameters to detect whether a customer will default loan or not.

Table Contents

• Analysing Methodology
• Technologies Used
• Conslusions
• Contributors

Analysing Methodology

1. Data Understanding : Working with the Data Dictionary and getting knowledge of all the columns and their domain specific uses .
2. Data Cleaning : Removing the null valued columns, unnecessary variables and checking the null value percentage and removing the respective rows.
3. Univariate Analysis : Analyzing each column, plotting the distributions of each column.
4. Segmented Univariate Analysis : Analyzing the continuous data columns with respect to the categorical column .
5. Bivariate Analysis : Analyzing the two variables behavior like term and loan status with respect to loan amount.
6. Recommendations : Analyzing all plots and recommendations for reducing the loss of business by detecting columns best which contribute to loan defaulters.

Technologies Used

1. pandas library for handling datasets
2. numpy library for handling series
3. seaborn library for better graphic graph plots
4. matplotlib library for graph plots

Conclusions

1. Surprising number of charged offs belonged to category “Verified” for “verification_status” and the huge number of “Not_verified” status indicates a major need to 2. revamp the verification process being followed.
2. Public bankruptcy was a strong indicator of default
3. High DTI should be a key deciding factor for lending.
4. Purpose of loan like “education", "small business” are likely to default

A Detailed Analysis and recommendations are including in the pdf attached

Contributors:

• Aniket Singla
• Smita Behera