Feature Selection by Combining Fisher Criterion and Principal Feature Analysis

Feature selection is one of the most important issues in the fields such as data mining, pattern recognition and machine learning. In this study, a new feature selection approach that combines the Fisher criterion and principal feature analysis (PFA) is proposed in order to identify the important (relevant and irredundant) feature subset. The Fisher criterion is used to remove features that are noisy or irrelevant, and then PFA is used to choose a subset of principal features. The proposed approach was evaluated in pattern classification on five publicly available datasets. The experimental results show that the proposed approach can largely reduce the feature dimensionality with little loss of classification accuracy.