A Meta-Analysis of Classification Methods

1. Aim of the project Supervised classification has a substantial history, but has recently progressed dramatically as a result of the advent of the computer, and is now considered as one of the main components of the relatively new field of data-mining. Numerous methods have been designed in the past twenty years or so, but by different communities who had different approaches and interests. Comparative studies have thus been carried out to assess the relative merits of the methods. Perhaps paradoxically, however, there is still little understanding of the circumstances under which certain methods will perform better than others. Providing some elements of such an understanding was our initial and ideal goal.