This project is aimed at building a ML model to predict whether an applicant is approved for a loan. The project includes data preprocessing, exploratory data analysis (EDA), model training, and deployment using Docker.

The csv file includes 58645 records with 13 columns:

• id
• person_age
• person_income
• person_home_ownership
• person_emp_length
• loan_intent
• loan_grade
• loan_amnt
• loan_int_rate
• loan_percent_income
• cb_person_default_on_file
• cb_person_cred_hist_length
• loan_status

The project's EDA and model training is documented in the Notebook

Model is trained through running this script. The trained model is saved in xgboost_model.pkl and dv.pkl

The notebook is developed using python 3.9.20 with miniconda. The entire dependencies is extracted to environment.yml file.

The training script can be run with python + venv:

python -m venv my-venv
source my-venv/bin/activate
pip install --no-cache-dir -r requirements.txt

python train.py

The model is deployed with Flask in app.py and using docker for containerization.

Build docker image:

docker build -t loan-approval-proj .

Run the image:

docker run --rm -p 5000:5000 loan-approval-proj

The model result can be tested directly with predict.py

Or by calling the containerized service with:

curl -i -X POST http://localhost:5000/predict -H 'Content-Type: application/json' -d '{"loan_grade": "C", "person_home_ownership": "RENT", "loan_percent_income": 0.07, "loan_intent": "MEDICAL", "person_income": 30000, "person_emp_length": 0.0}'