Time Series Churn Prediction

Problem Statement

A telecom company was experiencing high customer churn. The goal was to build a model to predict which customers were most likely to churn.

Methodology

Utilized time-series analysis on customer usage data. Developed a machine learning model using Logistic Regression and Gradient Boosting.

Key Results

The model achieved 85% accuracy in predicting churn. Delivered a dashboard visualizing key churn indicators.

Business Implications

Enabled proactive customer retention campaigns targeting at-risk customers, projected to reduce churn by 15%.

Tools Used