Abstract

In the contemporary financial landscape, predicting customer transactions plays a crucial role in enhancing customer service, personalizing marketing strategies, and improving operational efficiency. This research paper delves into the prediction of customer transactions using machine learning. Various machine learning techniques have been employed in previous research to predict customer transactions. Utilizing the anonymized Customer Transaction Prediction dataset, this study undertakes a comprehensive data analysis, rigorous feature engineering, and model training. The primary aim is to predict the likelihood of a customer making a specific transaction in the future. The methodology encompasses various data visualization techniques, statistical analyses, and model evaluation metrics to ensure robust and accurate predictions. Our findings demonstrate the effectiveness of the LightGBM model in handling large-scale datasets with numerous features, achieving a competitive AUC score.