User pattern based online fraud detection and prevention using big data analytics and self organizing maps

Online banking is the one most common service availed by almost all banking customers in the current era. Every second the banking organization, generate enormous amount of valuable data from their customers and their transactions. These valuable data need to be saved and analysed effectively using big data analytic techniques so as to get the necessary insights for the banking organizations. In today's market trend, analysing large data sets comprising of variety of data is of high importance to discover hidden patterns, market tendencies, customer likings and other business insights. The purpose of this research paper is to suggest a machine learning and big data analytics technique to detect and prevent any fraudulent online transactions. The model allows storage of the huge volume of online transaction data, which is then cleaned and features were extracted and reduced using the principal component analysis method. The reduced features are used to train the machine learning model, which is used to identify and recognize the user patterns related to e-transactions. Any e-transactions carried out by the user, the algorithm first checks for the matching user patterns, if there is a match, then the transaction will be successful otherwise the transaction will be reported as fraudulent. Thus the stored patterns created by the self-organizing map algorithm will detect and prevent the unauthorized access on banking transactions.