Predicting Loan Repayment: A Deep Learning Journey with Lending Club Data

This project focuses on analyzing Lending Club loan data, building a machine learning model to predict loan repayment, and evaluating its performance. The analysis involves data cleaning, exploratory data analysis (EDA), feature engineering, and the creation of a neural network model using TensorFlow.

About Data:

LendingClub is a US peer-to-peer lending company, headquartered in San Francisco, California. It was the first peer-to-peer lender to register its offerings as securities with the Securities and Exchange Commission (SEC), and to offer loan trading on a secondary market. LendingClub is the world's largest peer-to-peer lending platform.

• Loan Data: lending_club_loan_data.csv
• Data Description: lending_club_info.csv

Our Goal:

Given historical data on loans given out with information on whether or not the borrower defaulted (charge-off), can we build a model that can predict wether or nor a borrower will pay back their loan? This way in the future when we get a new potential customer we can assess whether or not they are likely to pay back the loan. Keep in mind classification metrics when evaluating the performance of your model!

Table of Contents:

1. Data Description Setup
2. Importing Libraries & Loading Data
3. Exploratory Data Analysis
4. Data Cleaning
5. Feature Engineering & Handling Categorical Variables
6. Train Test Split & Data Normalization
7. Creating & Training the Model
8. Evaluating Model Performance
9. Predicting Output
10. Conclusion

Data Description Setup

To understand the data, we start by loading a data dictionary that provides descriptions for each feature. This helps us gain insights into what each feature represents and its potential impact on the target variable.

Importing Libraries and Loading Data

We import the necessary libraries, including -

• pandas for data manipulation.
• numpy for numerical operations.
• matplotlib and seaborn for data visualization.
• tensorflow for building the deep learning model.

We load the Lending Club loan data into a pandas DataFrame. This data includes various features related to loan applications, such as loan amount, interest rate, employment length, and loan status.

Exploratory Data Analysis

Target Variable Analysis

We analyze the target variable, 'loan_status', to understand the distribution of fully paid and charged off loans.

Feature Distribution

We visualize the distribution of key features, such as loan amount, using histograms. This helps us understand the range and frequency of loan amounts in the dataset.

Correlation Analysis

We examine the correlations between continuous features using a heatmap. This helps us identify highly correlated features that may impact our model.

Data Cleaning

Handling Missing Values

We identify and handle missing values in the dataset. For instance, we explore features like 'emp_title' and 'emp_length' to decide whether to impute or drop them.

Removing Irrelevant Features

We drop features that are not relevant or have too many unique values, such as 'title' and 'emp_title'.

Feature Engineering and Handling Categorical Variables

• We engineer new features, such as 'loan_repaid', which converts the loan status into a numeric binary variable for modeling purposes.

• We convert categorical variables into dummy/indicator variables to make them suitable for machine learning algorithms. For example, we transform the 'home_ownership' feature into dummy variables representing different ownership statuses.

Train Test Split and Data Normalization

• We split the data into training and testing sets to evaluate our model's performance on unseen data. This involves defining the feature matrix 'X' and target vector 'y', and then using a train/test split function.

• To ensure that all features contribute equally to the model, we normalize the feature data using MinMaxScaler. This scales the features to a range between 0 and 1.

Creating and Training the Model

• We build a sequential neural network model using TensorFlow. The model consists of multiple layers, including input, hidden, and output layers. Dropout layers are added to prevent overfitting.

• We train the model on the training data and validate it on the testing data. The training process includes multiple epochs to optimize the model parameters.

Evaluating Model Performance

• We evaluate the model's performance using metrics like validation loss, classification report, and confusion matrix. These metrics help us understand the model's accuracy and precision.
• Plots training and validation loss curves to assess model learning behavior.

Predicting Output

We test the model by making predictions on new, unseen data. This involves selecting a random sample from the dataset and predicting whether the loan will be repaid.

Conclusion

This project demonstrates the application of machine learning to predict loan repayment status. By following a structured approach to data analysis, cleaning, and modeling, we can build a reliable model to assist financial institutions in assessing loan applications.