ABSTRACT
Thyroid disorders are widespread. Many activities in the body are controlled by thyroid hormones, such as calories and heartbeats. The thyroid is a type of gland that manages the energy system in the blood. The main cause of thyroid disease is a variation in hormone quantities in the human bloodstream. Some hormones, such as triiodothyronine (T3), thyroxin (T4) and thyroid stimulating hormone (TSH), have valuable quantities within fixed ranges in human blood. The main objective of this chapter is to evaluate the statistical analysis of enhancing accuracy in hormones variations using various machine learning algorithms. In this chapter, we have used three ensemble meta classifiers “AdaBoostM1, bagging, and random subspace” with a Pearson correlation features selection technique. The reduced error pruning decision tree generates as candidate subsample for pruning and formatting in a leaf node, but reduced error pruning has a drawback in that it cannot manage error complexity and accuracy as an ensemble model. So we have tested the reduced error pruning tree in three different environments, AdaBosstM1, bagging, and random subsample. We analyzed these tree ensemble methods and found that sensitivity, Youden’s J and accuracy values were high for the bagging meta classifier ensemble method with reduced error pruning and calculated better values compared to the AdaBoostM1 and random subspace meta classifiers ensemble methods.
