An Experimental Comparative Study on Thyroid Disease Diagnosis Based on Feature Subset Selection and classification

126001-8989 IJECS-IJENS © February 201 2 IJENS I J E N S Abstract— In this study several methods of feature selectio n and classification for thyroid disease diagnosis, w hich is one of the most important classification problems, are proposed. Two common diseases of the thyroid gland, which release s thyroid hormones for regulating the rate of body’s metaboli sm, are hyperthyroidism and hypothyroidism. Classification of these thyroid diseases is a considerable task. An importa nt problem of pattern recognition is to extract or select feature set, which is included in the pre-processing stage. As a case stu dy, Sequential forward selection and sequential backward selection , which are two well-known heuristic schemes, are employed for feature selection. Another feature selection method conside red is genetic algorithm, the popular method for nonlinear optimization problems. Support vector machine is used as classif ier to separate the thyroid diseases. This study is based on two th yroid disease datasets. The first dataset is taken from UCI machi ne learning repository and the second one is the real data whic h has been gathered by the Intelligent System Laboratory of K. N.Toosi University of Technology from Imam Khomeini hospital.