Review of credit risk and credit scoring models based on computing paradigms in financial institutions

Modern financial credit-disbursing institutions are characterized by fairly complex processes that struggle to improve the accuracy and predictability of credit scoring models. A bewildering array of studies have proposed methodologies to adapt big data analytics to this problem. This paper offers a brief overview of major studies and compares techniques along the following five dimensions: expected response time, threshold of input data, accuracy of output, reliability and computational overhead.