Patients with thyroid disease (TD) boast continuously increasing because of excessive growth of thyroid
gland and its hormones. Automatic classification tools may reduce the burden on doctors. This paper
evaluates the selected algorithms for predicting thyroid disease diagnoses (TDD). The algorithms
considered here are regularization methods (RM) of machine learning algorithms (MLA). The analysis
report generated by the proposed work suggests the best algorithm for predicting the exact levels of
TDD. This work is a comparative study of MLA on UCI thyroid datasets (UCITD). The developed system
deals with RM i.e., ridge regression algorithm (RRA) & least absolute shrinkage and selection operator
algorithm (LASSO). The above algorithms personage produce at most 79% accuracy by RRA and 98.99%
accuracy by LASSO. Thus, this paper shows the importance of LASSO, along with an example for
parameter generation. The decisive factors (DF) also suggest the accuracy rate of LASSO is much better
when compared with RRA.