Loan Risk Predictor

The project uses machine learning to classify whether granting loan to a particular customer is safe or risky.

It uses XGBoost Classifier to make predictions on data and is deployed as a web app using Flask framework. The classifer has been optimized over the ROC AUC metric and gives a score of 0.86.

Structure of the Project

The contents of the repository are as follows

• app - Folder containing the flask app code.

  • static - folder containing images and CSS for the website.
  • templates - folder of HTML files that are rendered.
  • routes - python code file of the backend created using flask.
  • categories - python file containing list of categories for the categorical variables in the html form.
  • init - python file used to initialize a module. It creates the flask app object.

• Loan Defaulter Prediction - jupyter notebook that was used to carry out data analysis, feature engineering, and modelling of the machine learning classifier.

• my_pipeline - pickle file of the machine learning pipeline that is used to preprocess and make predicitions of the user input.

• Data - csv file of the data over which the model was trained and tested. Source - Kaggle

• run - python file that imports the flask app object from the app module and runs it on the local server, in the debug mode.

• requirements - a .txt file that lists all the required packages for running the flask app and viweing the jupyter notebook.

Installing the Project

1. Download all the files in the repository.

2. Install all the required packages in your python environment listed in requirements.txt file.

   • Using pip :

   pip install -r requirements.txt

   • Using conda :

   conda create --name <env_name> --file requirements.txt

3. Open terminal/anaconda prompt and activate the environment with required packages.

4. Enter the root directory.

Running the Flask app on the local server

To run the app, run the following command on terminal :

python run.py

If everything runs correctly, the output will be

 * Serving Flask app "app" (lazy loading)
 * Environment: production
   WARNING: This is a development server. Do not use it in a production deployment.
   Use a production WSGI server instead.
 * Debug mode: on
 * Restarting with stat
 * Debugger is active!
 * Debugger PIN: 181-754-048
 * Running on http://127.0.0.1:5000/ (Press CTRL+C to quit)

Copy the url and open it on a web browser. You can now see the website running.

Try it out by filling the form with customer details and getting the risk predictions!