I have decided to analyse one of the datasets in the options, Prosper Loan Data. It includes data on 100k+ loans, with extensive information and various details on them. By looking at the Data Dictionary I decided to focus on 12 features of these loans, namely:
- ListingNumber: The number that uniquely identifies the listing to the public as displayed on the website.
- ListingCreationDate: The date the listing was created.
- CreditGrade: The Credit rating that was assigned at the time the listing went live. Applicable for listings pre-2009 period and will only be populated for those listings.
- Term: The length of the loan expressed in months.
- LoanStatus: The current status of the loan: Cancelled, Chargedoff, Completed, Current, Defaulted, FinalPaymentInProgress, PastDue. The PastDue status will be accompanied by a delinquency bucket.
- ClosedDate: Closed date is applicable for Cancelled, Completed, Chargedoff and Defaulted loan statuses.
- ProsperRating (numeric): The Prosper Rating assigned at the time the listing was created (applicable for loans originated after July 2009):
- 0 - N/A
- 1 - HR
- 2 - E
- 3 - D
- 4 - C
- 5 - B
- 6 - A
- 7 - AA
- ProsperScore: A custom risk score built using historical Prosper data. The score ranges from 1-10, with 10 being the best, or lowest risk score. Applicable for loans originated after July 2009.
- ListingCategory: The category of the listing that the borrower selected when posting their listing:
- 0 - Not Available
- 1 - Debt Consolidation
- 2 - Home Improvement
- 3 - Business
- 4 - Personal Loan
- 5 - Student Use
- 6 - Auto
- 7 - Other
- 8 - Baby&Adoption
- 9 - Boat
- 10 - Cosmetic Procedure
- 11 - Engagement Ring
- 12 - Green Loans
- 13 - Household Expenses
- 14 - Large Purchases
- 15 - Medical/Dental
- 16 - Motorcycle
- 17 - RV
- 18 - Taxes
- 19 - Vacation
- 20 - Wedding Loans
- Occupation: The Occupation selected by the Borrower at the time they created the listing.
- EmploymentStatus: The employment status of the borrower at the time they posted the listing.
- IsBorrowerHomeowner: A Borrower will be classified as a homowner if they have a mortgage on their credit profile or provide documentation confirming they are a homeowner.
Cleaning: The dataset is made of 112481 rows and 12 columns. It's a subset of the original ProsperLoanData dataset as I want to focus on fewer variables. I have also dropped a few rows containing an invalid value for the Prosper Score and rows having N/A for both Prosper Rating and Credit Grade.
The univariate distributions are all generally as expected, with some interesting insights that we will investigate further in the next step.
- Term of the loan:
- The more recent years have seen an increasing trend in number of loans being created and closed.
- Most of the loans have a 3 year term.
- The duration of the loans follows a bimodal distribution with peaks at the 1 year and 3 years points.
- Credit Rating/Score:
- By looking at the Credit Grade, Prosper Rating and Prosper Score, we found out that most of the borrowers have a credit score lying in the middle, ie the credit score roughly follows a normal distribution.
- Loan features:
- Most of the loans are current or completed, with only a very small portion being with status past due, defaulted or charged off.
- Among the reasons for which loans have been taken out, the most common are: Debt Consolidation, Home improvement, Business. A quarter of the reasons were classified as "Other".
- Borrower's features:
- As expected, most of the borrowers (ca 95%) are employed.
- An interesting fact is that the population is split in half between borrowers owning a house and not owning a house.
- The borrower's occupation is not detailed enough for most of the population, we can therefore avoid analysing this variable in the next stage.
For the presentation, I will mention the 50-50 split of homeowners/non-homeowner and show trends for Term, Duration, Credit Score, Loan Status.
Some insights from the bivariate exploration:
- Prosper Score and Prosper Rating are highly correlated variables.
- People with a lower Prosper score or Credit Rating tend to have a higher chance to have their loans charge off or default. This also tells us that the rating is a good and useful indicator for Prosper, which is always good news!
- There is a peak of homeowners where rating is AA/A or 6/7 and the amount is decreasing with lower credit ratings.
- There is a smaller proportion of homeowners in the sample of defaulted and charged off loans, compared to the full population. One other feature of interest was the duration of the loans. Before 2009, only 3Y loans were offered, which obviously makes the analysis by loan term more complex. Also, a lot of defaulted/charged off loans were closed in 2009.
Among the completed loans, ie excluding current and past due, generally speaking 1Y and 3Y loans lasted the whole term, whereas 5Y loans were completed earlier.
However, this might due to the fact that the first 5Y loan was created in 2010 and the dataset is until 2014 so we would need data for the following years as well to make conclusions on whether 5Y loans get repaid earlier or not.
Another interesting insight is that, even though the most common loan categories remain the same across the different loan terms, higher the term, higher is the proportion of loans for Debt Consolidation.
For the presentation, I will show the following relations:
- Loan Status vs Rating
- Homeowner vs Rating
- Duration vs Term
- Duration vs Loan Status
- Term vs Loan Status
- Prosper Score vs Rating
I will also mention that Prosper Score and Prosper Rating are two highly correlated variables. As these two are used after 2009, I have created a new Rating variable that picks up Credit Grade where populated and, otherwise, Prosper Rating and use these for the analyses.
During the last phase of this exploratory analysis, it became obvious that owning a home and credit rating are good indicators of loan default (or writing off). For instance, 61% of the HR loans defaulted/ got charged off and the majority of these were non-homeowners.
I expected defaulted/ charged off loans to be closed mostly after the final term date. However, 18% of the overall population of loans is closed after the term final date whereas this is true for only 4% of the defaulted/ charged off loans. This is a sign that loans that are meant to default get identified generally earlier than the final date, which is a good thing for Prosper.
Two plots will be shown for the presentation:
- Term vs Duration vs Loan Status
- Credit Rating vs Homeowner vs Loan Status
The main thread of my presentation and analysis is the focus on indicators for loan default (or writing off). I have looked at the difference between agreed term and actual duration, at credit rating and at whether borrowers own a home. These variables were analysed individually but also in relation to one another.