Machine Learning Based Fantasy Cricket Prediction System with Chatbot Integration

Shubham Yashwant Jadhav
Mentor: Mrs. Pallavi Sakalley

Department of Artificial Intelligence, Parul University

Random Forest models on IPL player data; (ii) integration of prediction results into a real-time chatbot interface using Flask REST APIs; and (iii) an end-to-end system architecture accessible to non-technical fantasy cricket users.

Abstract — Fantasy cricket platforms such as Dream11 rely heavily on user intuition for team selection, often lacking data-driven decision support. This paper presents a machine learning-based system for predicting IPL player performance and match outcomes, integrated with a conversational chatbot interface for real-time user guidance. A dataset comprising 4,200 player records across IPL seasons 2018-2023 was utilized, with features including batting averages, strike rates, bowling economy, venue statistics, and recent form (last 5 matches). Three models were evaluated: Decision Tree (71.4% accuracy, F1: 70.8%), Naive Bayes (68.9% accuracy, F1: 67.5%), and Random Forest (83.2% accuracy, F1: 82.7%, AUC-ROC: 86.4%). The proposed system bridges the gap between complex ML predictions and non-technical users through a React.js and Flask-based architecture. Results confirm the superiority of ensemble learning for sports prediction tasks.Index Terms — Machine Learning, Fantasy Cricket, Prediction System, Chatbot Integration, Random Forest, Ensemble Learning, IPL Analytics, Decision Tree, Naive Bayes, Flask, React.js, Sports Analytics