Machine Learning Models for Relevant Feature Identification and Classification of Thyroid Data

Abstract

Inappropriate creation of thyroid glands turn into prime subject of concern amongst Indian women. The two key chaos of thyroid which ought to be taken care at the earliest are hypothyroidism and hyperthyroidism. The improper secretion of thyroid may leads to obesity, fertility related problems, feeling depressed, etc,. Most of the thyroid problems can be managed if it has been properly treated. In recent years a number of models have been developed to investigate the thyroid-disorder. The laboratory tests are conducted to find the levels of the hormones and some of the physical examinations are used to identify the presences of thyroid. These examinations and test results are taken as the feature for developing the model. The feature importance can promote the performance of ML algorithms. The core intention of this study is to improve the classification performance by identifying the relevant features before classification. In this work, random-forest model is considered for identifying the important features and KNN algorithm is implemented for multiclass-classification to envisage the kind of thyroid chaos. Applying KNN after the feature selection improves the prediction accuracy. The developed model can be used to predict the presence of thyroid so that it can be treated accordingly.