Sparkify Customer Churn Prediction

Table of Contents

1. Installation
2. Project Motivation and Description
3. File Descriptions
4. Authors and Acknowledgements

Installation

Following libraries must be installed for running the notebook:

- pyspark
- pandas
- matplotlib
- seaborn
- user_agents (pip install pyyaml ua-parser user-agents)

Project Motivation and Description

This project aims to predict customer churn for the fictional music streaming service Sparkify using PySpark.

In more detail, the goal is to identify users with a high churn risk by finding patterns in the behaviour of already churned users.

For this project a small subset of the user log data has been provided to locally build a pyspark model that can be scaled on a much bigger dataset on a distributed spark cluster.

The key parts of the project are:

• Exploratory data analysis to get familiar with the data and identify potential features
• Feature engineering (based on the previous analysis) that provides proper input for the machine learning model, leveraging sparks pipeline concept
• Selection, training, evaluation and optimization of a classification model
• A blog post on medium that documents the project with a technical audience in mind

File Descriptions

• Sparkify.ipynb: The jupyter notebook containing the exploratory data analysis, feature engineering and model development
• mini_sparkify_event_data.json: The provided subset of the user log data from Sparkify

Authors and Acknowledgements

This project has been implemented as part of the Udacity Data Scientist Nanodegree program. The data has been provided by Udacity accordingly.