- Project Predict Customer Churn of ML DevOps Engineer Nanodegree Udacity
This project trains two classification machine learning model (randomforest and logistic regression) to predict the likelihood of bank customers to churn. Other than training, the raw notebook ( churn_notebook.ipynb
) is rewitten to PEP8 standard and scores a near 8.0/10 on pylint scale.
- The training and testing datasets are found in
data
- A quick intuition into the dataframe through plots can be found in
./images/results
- The trained models are saved in
./models
- A refactor and modularized version of
./churn_notebook.ipynb
is saved as./churn_library.py
- Create environment:
conda create --name churn_predict python=3.6
conda activate churn_predict
- Install packages:
conda install --file requirements.txt
- Run churn prediction:
python churn_library.py
- Test churn prediction:
python churn_script_logging_and_tests.py