renee127 / classifying_churn

Data Analysis Project (03/2022): Final project for Practicum by Yandex Data Science Bootcamp. This project uses machine learning techniques to develop a predictive model of customer churn with AUC-ROC > 0.92.

classifying_churn

Description

This data analysis project uses machine learning and data balancing techniques to create a predictive model higher than the target AUC-ROC (0.93 versus goal of > 0.88).

Project Overview

• Interconnect provides landline communication, internet, and several complimentary services.
• Clients select a monthly payment or yearly contracts.
• Interconnect wants a model for predicting the churn of clientele.
• If a user is forecast to leave, they will be offered promotional codes and special plan options.

Data Description

• Data is valid as of February 1, 2020.
• Four data files from different souces are provided.

1. contract.csv - contract information
2. personal.csv - the client's personal data
3. internet.csv - information about Interenet services
4. phone.csv - information about telephone services

• The files are linked by the customerID column which contains a unique code assigned to each client.
• The target feature is the EndDate column equal to No.

Project Goal

• Create a classification model that predicts if a customer will leave soon based on the data files supplied.
• AUC-ROC is the primary metric and needs to be at least 0.81, though greater than or equal to 0.88 is ideal.
• Accuracy will also be reported.

Dependencies

This project requires Python and the following Python libraries installed:

NumPy
Pandas
matplotlib
seaborn
math
time
functools
re
IPython.display
sklearn
catboost
lightgbm
xgboost
random
sys

You will also need to have software installed to run and execute an iPython Notebook.

Authors

Renee Raven

License

This project is licensed under the MIT License - see the LICENSE file for details