An Experimental Study on Hypothyroid Using Rotation Forest

Abstract

This paper majorly focuses on hypothyroid medical diseases caused by underactive thyroid glands. The dataset used for the study on hypothyroid is taken from UCI repository. Classification of this thyroid disease is a considerable task. An experimental study is carried out using rotation forest using features selection methods to achieve better accuracy. An important step to gain good accuracy is a pre-processing step, thus here two feature selection techniques are used. A filter method, Correlation features subset selection and wrappers method has helped in removing irrelevant as well as useless features from the data set. Fourteen different machine learning algorithms were tested on hypothyroid data set using rotation forest which successfully turned out giving positively improved results.