About

What I Built:
An end-to-end machine learning workflow on Zerve that predicts customer churn using historical telecom data.

Problem It Solves:
Helps businesses identify customers likely to churn in advance, enabling proactive retention strategies and reduced revenue loss.

Validation Approach:

Train-test split

Evaluation using Accuracy, Precision, Recall, and F1-score

Focused on churn-class performance

Results:

Accuracy: 93.37%

Churn F1-score: 0.88