Big Data Analytics for Tax Administration

Abstract

The problem of tax evasion is as old as taxes itself. Tax evasion causes several problems that affects the growth of a nation. In this paper, we present our work in controlling tax evasion by using big data analytics, Android applications, and information technology. We implemented this work for the commercial taxes department, government of Telangana, India. Here we developed a complete software framework for scrutiny of suspicious accounts. This system detects suspicious dealers using certain sensitive parameters and standardizes the process of scrutiny of accounts. We used sophisticated statistical and machine learning tools to predict suspicious dealers. To increase the compliance levels, we developed a regression model for identifying return defaulters and user-friendly Android applications to assist the officers in collecting the tax. The other aspect we explored is the detection and analysis of a tax evasion mechanism, known as circular trading, using advanced algorithmic and social-network analytic techniques.