End-to-end customer churn prediction pipeline using Spark.

A distributed approach for data exploration, neural net training prediction and scoring. The modules are intended to handle large datasets by only operating in batches. The specific problem studied here is related to predicting the churn ratio (i.e., the amount of customers that leave a service) from a telecommunications company.

Churn prediction using Dask distributed to predict churns.

Это end-to-end ML-пайплайн для предсказания оттока клиентов на больших данных, демонстрируя распределённые вычисления для портфолио. Фокус на scalability и MLOps. Задача проекта это бинарная классификация — предсказать, уйдёт ли клиент (churn) на основе исторических данных (транзакции, поведение, демография).

Customer Churn Prediction using Apache Spark - Scalable & Distributed Computing coursework

Team Members: Navaneethakannan K, Nandhagopal B

Big data churn prediction system built with PySpark, Spark SQL, Spark MLlib, and distributed batch processing.

A production-ready Telco Churn Prediction system with Scikit-learn pipelines, MLflow experiment tracking, PySpark for distributed processing, and Airflow for workflow orchestration.

Industry-level distributed churn prediction pipeline for telecom data using PySpark and Spark MLlib. Focused on scalable ETL, feature engineering, and ML modeling for enterprise datasets.

InSightX is a distributed churn prediction system for telecom data, using Spark, Kafka, and AWS EMR. It processes 10M+ records with real-time ingestion and achieves an 88% F1 score using Gradient Boosted Trees.

Most existing classification approaches assume the underlying training set is evenly distributed but many real-world classification problems have an imbalanced class distribution, such as rare disease identification, fraud detection, spam detection, churn prediction, electricity theft & pilferage etc.