An efficient money Laundery detection using Spatio temporal correlation based association rule pattern

The implementation of data mining techniques for money laundering fraud detection follow traditional information flow of data mining, which starts with feature selection followed by representation, data collection, management, and performance evaluation. Data mining methods have the potentiality for detecting money laundering fraud in banking as they use past cases of fraud to build models, which identify and detect the risk of fraud.With the increased volume of crime datasets and complexity of relationships between these kinds of data, several data mining approaches were presented that involved anomaly detection using principal component analysis and self organizing map. But nevertheless, with high dimension data, they pose serious issues. In this work, to handle high dimensional data with multi-clustering structure, an Efficient Association Rule Pattern based Money Laundering Detection (EARM-MLD) framework is developed. The association rule pattern mining in EARM-MLD framework consists of three major parts. The first part finds frequent large itemsets from banking rules which have support and confidence values more than a threshold number of times. This in turn reduces the time taken for detecting money laundering. The second part is to construct association rules based on spatio temporal model from those large itemsets to easily perform the detection operation and to integrate it with multi clustering algorithm with the objective of reducing the false positive rate. Finally, the multiclustering algorithm involves the set of money transfer group which fulfills the criteria such as row condition, gathering amounts of money to a single account with minimum set size. The multi cluster elements integrated with EARM-MLD framework are treated as suspected operations which operate in money laundering detection work.The money laundering detection in banking system using EARM is experimented on factors such as time for detecting money laundering,false positive rate, scalability, system efficiency ratio, fraud identification accuracy, number of transaction , number of money transfers. Experimental analysis shows that EARM-MLD framework is able to reduce the time for detecting money laundering by 40.14% and reduce the false positive rate by 18.55% compared to the state-of-the-art works.