Profiling and clustering methods for transaction profiling in BRT transaction

This paper works on fraud identification using transaction profiling in Bus Rapid Transit transaction. The research has purpose to deliver profiling information for fraud identification baseline. The data used by the research reach 22GB for 2 years transaction, which has 165 million records. The data process using MapReduce environment that placed in the 9 nodes Hadoop Cluster. The approach has implemented with two-way methods, there is value and amount based profiling. The profiling has been implemented for generating the profile of card, time, gate, and prepaid transaction. The research has strengthening the data process by defined the data transformation into common format, data mapping, selecting the attributes, and generate the value and amount. The works has shown that audit trail profile has resulted by profiling and clustering process from BRT transactions.