A system, method, and computer-readable medium for performing a payment tracking operation. Factors, that include input, features and attributes of various customers are received and processed. The processing includes imputing, deleting and converting the factors for further refinement for machine learning. The machine learning processes the factors using succeeding layers that include random forest to categorize customers, multivariate regression to determine which factors have the greatest effects on payment. In addition, a recurrent neural network approach is used to further refine the factors and provide a more accurate analysis.