Digital event profile filters based on cost sensitive support vector machine for fraud detection, risk rating or electronic transaction classification

摘要

A computer system includes a device having instructions to implement a training application programming interface that is configured to receive a first set of data that includes training observable events and a model builder engine that is configured to provide estimated outcomes of said training observable events using a kernel function, wherein said kernel function is configured by a plurality of parameters including a match reference vector and a mismatch measurement vector. The model builder engine is also configured to calculate values of said mismatch measurement vector to provide a calculated mismatch measurement vector calculate a distance between the value of said match reference vector and the value of said calculated mismatch measurement vector corresponding to each training observable event. The model builder engine is also configured to determine an accuracy of said kernel function in providing estimated outcomes of said training observable events.