Predict Customer Churn

Project Predict Customer Churn of ML DevOps Engineer Nanodegree Udacity

Project Description

This project is used to demonstrate a customer churn analysis using Clean Code best practices. This project is intended to demonstrate composing, testing, and logging an analysis in modular functions using PEP 8 coding standards.

Files and data description

The directory structure for the projects is as follows:

.
├── Guide.ipynb          # Provided by Udacity: Getting started and troubleshooting tips
├── churn_notebook.ipynb # Provided by Udacity: Contains the code to be refactored
├── churn_library.py     # Functions defined for the churn analysis
├── churn_script_logging_and_tests.py # Testing and logs for executing the churn analysis
├── constants.py # Constant values used in the script
├── README.md            # Provides project overview, and instructions to use the code
├── data                 # Data provided for the analysis
│   └── bank_data.csv
├── images               
│   ├── eda              # Store EDA image results
│   └── results          # Store model training/testing results
├── logs                 # Store logs
└── models               # Store models

The directory contains two primary submission artifacts from the project, the churn_library.py and churn_script_logging_and_tests.py files.

• The churn_library.py module contains functions composed to complete the churn analysis.
• The churn_script_logging_and_tests.py file is provided to demonstrate an example execution of the churn library functions, as well as a demonstration of logging and unit testings for the functions.

Dependencies

This project relies on the following dependencies

• scikit-learn
• joblib
• pandas
• numpy
• matplotlib
• seaborn

Running Files

In order to execute these files, an environment of python dependencies must be defined. The churn_script_logging_and_tests.py can be executed from the command line while the churn_library.py files provides a library of functions to be imported.

• The python environment can be resolved by executing python -m pip install -r requirements_py3.6.txt
• An example of logging/testing with this analysis library can be executed using ipython churn_script_logging_and_tests.py

Executing the churn_script_logging_and_tests.py files will write a log in to the ./logs directory and produced images to the ./images/.

The image locations, model parameters and other constants can be modified using the constants.py file.