Forecast Tax Non-Compliance in Rwanda Using Predictive Models

Rwanda’s economic development, aligned with global trends, depends heavily on tax revenue to finance critical infrastructure and public services including education, healthcare, public safety and transportation networks. These services are vital for achieving Rwanda’s Vision 2050 goals of sustainable growth and poverty reduction. However,tax compliance remains a significant challenge, with a substantial portion of the population, particularly among small-scale traders and rural taxpayers failing to file or pay taxes on time. This non-compliance limits the government’s ability to fund essential services and hinders Rwanda’s ambition to become middle-income economy.This study investigates the potential of machine learning models to predict tax non-compliance using historical taxpayer data from the Rwanda Revenue Authority(RRA) covering 2018-2023. By leveraging regression analysis and advanced predictive models such as Logistic Regression, Random Forest, XG-Boost and Decision tree,the study aims to identify individuals or businesses at high risks of failing to file or pay taxes on time. Additionally, it seeks to pinpoint key predictors of non-compliance such as income levels, business size, sector and geographic location.