Once we identified, through analysis of the data, a sub-set of attributes to include in our data-mining, our next step was to rank those attributes in relation to the dataset's dependant variable, CardType. This ranking was performed with the InfoGainAttributeEval tool in the Weka toolset.

Based on the output of the Weka system's ranking tool, we were able to reduce the number of attributes we used for our evaluation from twenty-six (26) to only nine (9). The attributes that were retained for our data evaluation can be seen here.