FRISS DATAMINING
The Friss Data mining module is deployed when a reasonable data set of proven fraud cases has been built up. The data mining module is used to deepen this and to generate new fraud profiles on the basis of historic data.Data mining is used to find significant differences and patterns between two sets of data, in this case the set ‘proven fraudsters’ and ‘proven non-fraudsters’. Such an analysis can be carried out if this set is large enough and there is also sufficient other information available about these cases. It then becomes clear from this analysis which variables or combinations of variables have significant different values between the sets. From this it could, for example, be clear that the colour of the eyes of the driver is relevant. The data mining models are then added to the detection process. The objectives can be aimed at improving the success level, searching for new links on the basis of existing fraud cases and determining the added value of new sources of data.
The data mining model can be fed with data from the Friss Fraud database or a data warehouse filled with the data from the Friss Fraud database. This latter database contains all data used for the assessment.
