Revue d'économie et de statistique appliquée
Volume 22, Numéro 2, Pages 79-97
2025-12-09

Predictive Models For Customer Churn In Ride-hailing: A Machine Learning Approach



Authors : Smail Mohamed Yacine . Asri Ayoub .

Abstract

This study investigates the application of advanced machine learning techniques for customer churn prediction within the ride-hailing industry, leveraging real-world transactional data from a regional service in Algiers. Customer churn is explicitly defined as the absence of rides within a six-month period following the evaluation period. The comprehensive methodology encompasses robust data preprocessing, which benefited from the naturally balanced distribution of the target variable, thereby simplifying the modeling pipeline. A systematic evaluation of multiple classification algorithms, including XGBoost, Random Forest, Logistic Regression, LightGBM, and Decision Tree, was conducted. A key aspect of this approach was the rigorous hyperparameter optimization using Grid Search, Bayesian Optimization, and Simulated Annealing, all implemented in R. The findings demonstrate that LightGBM, particularly when tuned with Simulated Annealing, achieves superior predictive performance (Accuracy: 0.762, Recall: 0.781, F1-Score: 0.779), effectively identifying potential churners. This research provides actionable information for ride-hailing companies to enhance customer retention strategies by focusing on critical behavioral patterns identified through feature importance analysis. The technical rigor, evidenced by the meticulous preprocessing and advanced optimization, culminates in a solution that offers direct business utility by enabling targeted interventions for customer retention.

Keywords

Customer Churn Prediction ; Ride Hailing ; Machine Learning ; Hyperparameter Tuning ; Lift Analysis

Les articles similaires

خدمات النقل عبر التطبيقات الالكترونية بين غياب الاطار التشريعي و تراكم الاشكالات القانونية Ride-hailing Services Via Electronic Applications: Between The Absence Of A Legislative Framework And The Accumulation Of Legal Issues

سعادة فاطمة الزهرة . 
ص 683-712.

Machine Learning Based Customer Churn Prediction: A Bibliometric Analysis Of Techniques, Key Methodologies, And Sectoral Trends (2011-2025)

Yahiaoui Nour El Houda . 
pages 135-158.

The Role Of Machine Learning Techniques In Improving Supply Chain Management: A Comparative Analysis Of Predictive Models With An Application To Wheat Consumption In Algeria

Kahoui Hacen .  Sahed Abdelkeder .  Chekouri Sidi Mohammed . 
pages 96-113.

Analysis And Development Of A Predictive Outlook Model For Negara Bank Of Indonesia Stock Using Machine Learning Data Training (python)

Gherbali Abdelhamid .  Dekkiche Djamal .  Machou Mustefa . 
pages 503-514.

An Approach Based On Multi Agents System And Genetic Algorithm For A Dynamic Dial A Ride Problem

Bensouilah Oussama .  Lekadir Ouiza .  Boukredera Djamila . 
pages 120-120.

Formats de citation