In this example, we've continued the scenario from the previous chapter to find out the
bestdata mining model that generates better results for the problem (which was finding
prospective bike buyersfor the new product). The best mining model would be the one that
generates results that are closest to the test dataset.
As a data mining developer, you are required to train multiple algorithms with the existing
dataset. The reason for using multiple algorithms is that different mining algorithms
generate patterns differently and give results differently. We will use multiple mining models
to find the best results for the defined problem, so we need to find the best algorithm that
produces the best results compared to the test dataset. After creating multiple algorithms
and training them, you can use a set of mining accuracy charts to figure out which algorithm
performs best compared to the real data in the test dataset. In this example, we've added
two new mining models to the previous Target Mail Mining Structure: clustering and Naïve
Bayes (steps 1 to 4).