Tax evasion is pervasive in most of the nations. It impedes the economic progress of the nation. In this paper, we devise a technique that detects groups of people who commit evasion in the goods and services tax by manipulating their sales and (or) purchases with the motivation to reduce their tax liability. In particular, we perform clustering analysis on certain sensitive parameters pertaining to each of the dealers. Then by analyzing the number of dealers across the different clusters, we identified the suspicious set of dealers who are further investigated by looking at other sensitive parameters. This work is designed and implemented for the Government of Telangana, India. © 2019 IEEE.