Towards an evolutionary algorithm: a comparison of two feature selection algorithms

In order to deal with a large number of attributes, probabilistic feature selection algorithms have been proposed. Pure random walk entails mediocre performance in terms of search time. Introducing adaptiveness into a probabilistic algorithm can lead to a more focused search that results in a better search time. We compare two algorithms in search of an efficient but not myopic algorithm for feature selection. Based on the comparative study, we suggest some ways of improvement towards an evolutionary feature selection algorithm for data mining