Hetul S. Shah
Abstract: It is commonly agreed that AR (account receivable) is most valuable asset of any business firm. It can be a source of financial difficulties for firm when they are not efficiently managed and under-performing. So, it is important to identify data pattern in AR and get meaningful insight from AR data. This paper demonstrates how supervised machine learning can help to build model to predict payment outcome of invoices which are yet not paid (Open) based on historical data. Proposed method can predict with high accuracy that in which age bucket (i.e. On time, 1-30, 31-60,.., Over 150) the invoices will be paid. This method is implemented in the context of real-world AR data. Finally, simulation results are shown which proves that this approach can give high accuracy and save significant amount of collection time.
Keywords: Predictive Modeling, Accounts Receivable, Supervised Classification, AdaBoosted Decision Trees