Abstract:
"The increasing rate at which bank loses its profitability and
liquidity due to various reasons which contribute to loan fraud
is alarmingly high. This eventually poses a threat on the asset
quality, survival of banks and the whole economy of the
country. Extensive research and literature findings revealed
that increasing Non-performing Assets in Sri Lanka can be
due to poor credit appraisal methods, poor management of
credit facilities, inappropriate structuring, and wrong selection
of borrower by the bank and mis-management of funds, poor
equity contribution, diversion of funds and poor managerial
knowledge by the customers and other environmental causes
such as government policies, taxation laws and changes in
consumer behavior. Thus, this research combines historical
loan data based on machine learning to forecast and predict
fraud by discovering hidden patterns and enable banks to take
timely actions to the Non-Performing Assets."
