Applying Machine Learning to Detect Periodicity in Transactional Banking Data

This paper analyses the effectiveness of applying machine learning algorithms to detect periodicity in transactional banking data. The study includes a review of the dataset and the necessary processing techniques to be applied for using the algorithms. The research design is detailed with the aim of achieving the best results efficiently. The results obtained are explained by analyzing the impact of the correct selection of the algorithm's parameters, data processing methods, and comparing the different algorithms. The main contribution of this work is addressing the detection of periodicity in transactional banking data, a problem that has not been widely explored despite the extensive use of machine learning algorithms in finance. The results obtained in this study can have an influence in the personal finances of end users, because it will enable the development of personalized financial services and products, as well as intelligent recommendation systems that, through the forecasting of future income and expenses, help them improve their personal financial management. The research concludes with a review of the results and recommendations for future projects to consolidate findings and develop new recommendations based on detected patterns.