Thyroid Disorder Classification using Machine Learning

Machine learning algorithms and data mining techniques are crucial for handling the enormous amount of data and information that exists today, especially in the health system. Nowadays thyroid disorder can lead to a number of health issues according to the level of disorder. In order to improve thyroid disorder procurement, the current study focuses on categorizing thyroid disorders using machine learning approaches. This study’s major goal is to divide this condition into eight groups: hyperthyroidism, hypothyroidism, normal and five other classes. This paper proposes a Multi-layer Neural Network model that classifies the thyroid disorder data set using advanced machine learning techniques, customized loss functions, and feature extraction. For the better classification of disease proposed model utilizes eight class classifiers. In proposed model, different variants of model based on Tensor Flow functional API using exotic feature engineering techniques was applied, its performance with parametric models like SVM and over the data set of 9172 samples with 30 features are compared.