The Analysis of GCFS Algorithm in Medical Data Processing and Mining

摘要

Feature selection plays a significant part in medical data processing and mining, it can reduce the dimensionalities of datasets and enhance the performance of the classifiers, and it is also helpful to clinical decision support to a great extent. At present, the clinical decision support is more performed by physicians subjectively based on clinical knowledge, which may hinder the diagnosis and treatment. This paper mainly outlines the performance of GCFS (Genetic Correlation-based Feature Selection) algorithm in the processing and mining procedure of medical data, and medical UCI datasets are employed as the studied materials for proving the improvement of feature selection in data classification. Compared with the algorithms of CFS and GA (Genetic Algorithm), ensemble learning methods are employed as the testing classifiers, and the results show GCFS algorithm almost improves the performances of the testing classifiers better than CFS and GA.