Ensemble Deep Learning for Cricket Score Prediction: Integrating CNN, LSTM, and DNN for Enhanced Accuracy

Cricket score prediction is a complex problem influenced by dynamic factors such as wickets lost, overs remaining, and run rate. These factors are critical because they directly affect batting strategies, scoring potential, and match outcomes. This study presents an advanced machine learning model that integrates Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Deep Neural Networks (DNN) through an ensemble learning approach to improve prediction accuracy. CNN captures spatial dependencies in match data. LSTM models sequential dependencies across overs, while DNN extracts complex feature interactions. The dataset comprises past ODI match records, including ball-by-ball statistics and contextual conditions. The ensemble model leverages the strengths of each architecture, ensuring robust predictions with minimal error. Experimental results show superior performance compared to traditional regression and standalone deep learning models. This research supports strategic decision-making for teams, broadcasters, and analysts by offering a data-driven approach to score forecasting. Future work will focus on hyperparameter optimization and expanding datasets for broader generalization across different match conditions.