AWSum - Data mining for insight

Many classifiers achieve high levels of accuracy but have limited use in real world problems because they provide little insight into data sets, are difficult to interpret and require expertise to use. In areas such as health informatics not only do analysts require accurate classifications but they also want some insight into the influences on the classification. This can then be used to direct research and formulate interventions. This research investigates the practical applications of Automated Weighted Sum, (AWSum), a classifier that gives accuracy comparable to other techniques whist providing insight into the data. AWSum achieves this by calculating a weight for each feature value that represents its influence on the class value. The merits of AWSum in classification and insight are tested on a Cystic Fibrosis dataset with positive results. © 2008 Springer-Verlag Berlin Heidelberg.