Brandon Kanyi Mwangi Projocet is the * syriatel.ipynb
Build a classifier to predict whether a customer will ("soon") stop doing business with SyriaTel, a telecommunications company. This is a binary classification problem.
Most naturally, your audience here would be the telecom business itself, interested in reducing how much money is lost because of customers who don't stick around very long. The question you can ask is: are there any predictable patterns here?
Slides
Model Performance: Through metrics such as accuracy, precision, recall, F1-score, and AUC-ROC curves, we evaluated and compared different models. The tuned Decision Tree model achieved a high accuracy (90.6%) and a balanced performance in predicting churn, making it an effective tool for the business.
Feature Importance: The feature importance plot provided insights into which factors are most influential in predicting customer churn. This information can be used by the business to focus on the key drivers of churn and implement targeted interventions.
Model Interpretation: The decision tree model is relatively interpretable, allowing the business to understand the logic behind predictions and take actionable steps based on this understanding.
The predictive model we developed can accurately forecast customer churn based on various customer attributes and behaviors. By leveraging this model, SyriaTel can:
Target At-Risk Customers: Identify customers at high risk of churn and take proactive steps to retain them, such as offering personalized incentives or improving customer service.
Optimize Marketing Strategies: Focus marketing efforts on features and services that reduce churn, based on the insights gained from feature importance.
Enhance Customer Experience: Understand common factors leading to churn and address underlying issues, such as billing disputes, service dissatisfaction, or lack of engagement.