Cricket Player Performance Analysis Using Deep Learning

Cricket is a sport where data-driven strategies play an essential role in enhancing player performance and game analytics. Traditionally, performance analysis has been performed manually, which requires analysts to review match footage and statistics, which is time-consuming and prone to human error. This study proposes an automated deep learning-based approach for analyzing cricket player performance by extracting match statistics and predicting shot outcomes. The research is divided into two main phases: (1) Data Extraction and Preprocessing, where unstructured match reports in PDF format are converted into structured CSV format using Node.js, Express.js, pdf-parse, json2csv, and MongoDB; and (2) Deep Learning Model Development, where a Deep Feedforward Neural Network (DFNN) is trained using extracted features such as ball speed, swing, line, length, and shot type to predict shot outcomes (Middled, Edged, or Missed). The proposed model, developed using TensorFlow and Keras, achieves an accuracy of 93.84 percentage in predicting shot outcomes. The evaluation metrics, including precision, recall, and the F1 score, confirm the reliability of the model. This research demonstrates the potential of artificial intelligence in sports analytics, enabling coaches and analysts to make datadriven decisions. Future improvements involve real-time data processing, video analysis, and expansion of the data set to improve the accuracy and robustness of the model.