In this paper, we explore machine learning (ML) methods to improve inflation forecasting in Brazil. An extensive out-of-sample forecasting exercise is designed with multiple horizons, a large database of 501 series, and 50 forecasting methods, including new ML techniques proposed here,...