Feature Selection Methods Applied to Severe Brain Damages Data
Wiesław Paja, Krzysztof Pancerz · Annals of Computer Science and Information Systems · 2017
Brain injuries seem to be one of the most widespread diseases.Hence, the main goal of our research was to investigate feature importance in the severe brain damages dataset according to the Glasgow Outcome Scale.This scale is recognized as one of several measures used to evaluate patients' functional ability as well as their conditions after applying brain damage therapy.The current approach is focused on an identification of a relevant subset of features with a similar influence on quality of classification models.According to the results gathered, about 12 from 42 descriptive features could be treated as important without the decrease of classification results.