Predict Customer Churn

• Project Predict Customer Churn of ML DevOps Engineer Nanodegree Udacity

Project Description

This project trains a classifier using data from kaggle. We want to predict customer churn with the bank_data.csv. The project trains a Random Forest classifier with different specifications as well as a Logistic Regression. The best performing Random Forest model is then used to calculate feature importance and other ML metrics.

File structure

Overall required folder structure:

• data
  • bank_data.csv
• images
  • eda
    • churn_distribution.png
    • customer_age_distribution.png
    • heatmap.png
    • marital_status_distribution.png
    • total_transaction_distribution.png
  • results
    • feature_importance.png
    • auc.png
• logs
  • churn_library.log
• models
  • logistic_model.pkl
  • rfc_model.pkl
• constants.py
• churn_library.py
• churn_script_logging_and_tests.py
• README.md

There is no requirements.txt, as this project only uses packages already defined in the notebook.

Description of python files

• constants.py: saves commonly used constants.
• churn_library.py: implements the class which implements the training and evaluation code.
• churn_script_logging_and_tests.py: implements the tests and logging. This file is required to actually run and train the model.

Running Files

To run the prediction, simply run

ipython churn_script_logging_and_tests_solution.py

Lint Code

Ppi install linting requirements

pip install pylint
pip install autopep8

Run pylint

pylint churn_library.py
pylint churn_script_logging_and_tests.py

Run autopep8

autopep8 --in-place --aggressive --aggressive churn_script_logging_and_tests.py
autopep8 --in-place --aggressive --aggressive churn_library.py