Published Paper Details:

Account receivables (AR) systems are a critical part of financial system of any business. Effective management of AR and financial performance of firms are positively correlated. Invoice to Cash (I2C) process, is a critical part of AR, which starts from the moment an invoice is created until the moment the customer's debt (payment) is settled or reconciled. Inefficient AR management delays cash realization, causing financial crunch of a given firm. In this context there have been attempts to address the issue of outstanding receivables through improvements in the collections strategy, specifically, through supervised learning to build models for predicting the payment outcomes of newly-created invoices, thus enabling customized collection actions tailored for each invoice or customer. Such models can predict if an invoice will be paid on time or not and can provide estimates of the magnitude of the delay. Predictive models use analytics to predict probability of event to happen. To effectively predict payment behavior of customers, the data that is generated in AR system has to be accurate in quality and interpretation, if not, may result in wrong predictions. Current predictive models which are based on business rules need to be data driven and the data has to be accurate, usage of proper algorithms, multiple models to make predictive models more comprehensive and thereby resulting in better business decisions. Improving approach of predictive models and data accuracy feeding into Predictive Model for Invoice scoring and usage of appropriate algorithms and comprehensive approach are dealt in this paper. The algorithm being proposed gives a statistical model, which uses business domain based variables