Thyroid is one of the most common disease found in people nowadays which occur due to disorder of thyroid gland that include hypothyroidism (inactive thyroid gland) and hyperthyroidism (hyperactive thyroid gland) that can take place at any age and in either sex. Therefore their prior diagnosis and detection is very crucial and helpful for the betterment of human life. Large amount of complex data is collected by the healthcare sector in order to identify hidden patterns for effective identification, detection and decision making. Data mining has become a current trend for achieving effective diagnostic result from massive dataset by classifying applicable and unique patterns in data. The aim of the paper is to present an extensive analysis of different classification techniques viz. naive Bayes, SVM, and K-nearest neighbour (K-NN) on the basis of dimensionality reduction for detection of thyroid disease. Results are provided to select best thyroid disease detection technique. The analysis reflected that K-NN is performing better than other classifier on the basis of various parameters. This analysis will help to identify the best algorithm for such diseases and give better preventive options in advance.