Data Analysis of Lending Club Data
The skills the authors demonstrated here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Based on data the "Lending Club" is a "peer to peer" lending company that provides various loans for individuals looking to finance personal loans, business loans, auto refinancing loans and medical loans. "Peer to peer" lending is a new form of lending that provides an avenue for lenders and borrowers to be matched online.
Founded in 2007, The Lending Club has now become one of America's largest online lending platforms. As a third party, the company makes money by selecting borrowers, smoothening transactions, and servicing the loans. The company not only allows its clients to borrow, but also provides an avenue for its other clients to invest in the loans being provided. Investors procure notes which match up to fractions of loans. The online platform has two primary business segments. The first segment is for the customer looking to borrow. The second segment is for customers looking to invest in the loans being created in the first segment.
As a result, this new form of lending makes it cheaper for borrowers to borrow money at lower interest rates, while allowing lenders to earn high er rates of returns than that currently provided at traditional banking institutions. Nevertheless, there is always a risk of the borrowers defaulting on the loans. As such, this R Shiny project is devoted to examining and finding some of the relevant data points to help those investing in the loans as well as help the company examine where it should focus its resources.
The loan data for this project was obtained through Kaggle and consists of loans issued from 2007 - 2015. The file was a matrix of about 890 thousand observations and 75 variables. However, this project focused on 6 primary variables which are: the loan amounts requested by the borrower, the loan status of the amounts borrowed, the length the borrowers have been employed at their respective places of occupation, the type of home ownership the clients have, the annual income of each borrower, and the respective credit grades assigned by "The Lending Club" to each borrower. A sample of 100 loans was used as the basis for the project.
Data visualization through the use of R Shiny was used to examine the different variables. The Lending Club assigns a credit grade to each borrower that ranges from an "A" which is the best rating to a "G" which is the worst rating. Furthermore, the loan status for each loan ranges from loans which are "paid off" or "current" to loans which are in "default" or have been "charged off".
The first visualization named "Explore Loan Variables and Grades" is used to explore the data consisted of a chart showing the "number of loans" on the Y-Axis with the "Credit Grades" on the X axis. R-Shiny gives the user to ability to examine how the customer variables: "loan purpose", "length of employment", "loan status" and "customer home ownership are distributed based on the credit grades. These variables can be selected by selecting them on "Fields to Analyze".
For example, when the "purpose of the loan" is selected, "credit card consolidation" stands out as the largest reason that funds were borrowed across all "credit grades".
With respect to "loan status" the chart shows that most loans are current across all credit grades. The largest amount of loans that have been charged off seems to belong to the loans in the "C" credit grades. This visualization is useful for "The Lending Club" and investors because it allows users to get a sense of what variables seem to be important when considering what loans to focus on.
Further analysis of the data sample was achieved through a second visualization named "Boxplot Summary" showing the various loan amounts on the Y axis with the corresponding field to be analyzed being shown across the X axis (length of customer employment, customer home ownership, loan purpose, and the credit grades). For example, when one considers the "loan purpose", the box plot shows "renewable energy" as the lowest purpose for a loan, whereas "credit card" and "debt consolidation" seem to be the most popular reason for loan requests.
This chart is useful because it gives the user of sense of what variables seem to stand out when exploring the portfolio of loans available by "The Lending Club".
A final visualization tool named "Explore Loan Status and Annual Income" is a chart showing the variable "loan status" on the "Y Axis" and the variable "Annual Income on the "X Axis". One can toggle through the three other variables (length of customer employment, customer home ownership, loan purpose) to analyze the different effects each variable on the different loans. For example, when one considers the "loan purpose", loans taken out for "home improvement" form a larger percentage of the amounts borrowed.
In conclusion, the examination of the data pertaining to "The Lending Club" provides a framework for the process by which the company may be able to explore and dissect its portfolio in order to determine the types of financing they would like to provide to its customers. For example, after examining the data, the company may find that it is more beneficial to provide credit to customers who have worked for a least 3 years, have a B tier credit grade and may be looking to consolidate their credit cards. This will help the company make more money in the long run.
Furthermore, the model will provide a framework to also allow customers to better determine the types of loans that they would like to invest in. For example, after examining the data, a customer could decide that they would like to invest in B tier customers, who rent their homes and have been employed for more than 10 years. The visualizations will thereby help investors make more money as well.