Enhanced Feature Selection Algorithm using Modified Fisher Criterion and Principal Feature Analysis

V. Arul Kumar, L. Arockiam


Dimensionality reduction is one of the key issues in various fields such as Data Mining, Machine Learning, Pattern Recognition, Image Retrieval, Text mining etc. This technique is used to reduce the dimensionality of features and improve the performance of learning algorithms. In general, dimensionality reduction is classified into two categories: feature selection and subspace learning. Recently, many researchers combined these two methods to improve the performance of the learning algorithm. In this paper, an enhanced feature selection algorithm is proposed namely, Modified Fisher Criterion Principal Feature Analysis (MFCPFA). The MFCPFA algorithm is developed by merging Modified Fisher Criterion (feature selection and subspace learning) and Principal Feature Analysis. To prove the effectiveness of proposed algorithm, the algorithm is tested with various datasets available in the UCI repository. The results show that the newly developed algorithm improves the classification and reduces the unwanted features.

Keywords: Feature selection, Fisher Criterion, Fisher Score, Linear Discriminant Analysis, and Principal Feature Analysis.

Full Text:


DOI: https://doi.org/10.26483/ijarcs.v3i4.1332


  • There are currently no refbacks.

Copyright (c) 2016 International Journal of Advanced Research in Computer Science