Tfb: An Open-source Machine Learning Library Designed For Time Series Researchers

The Time series Forecasting Benchmark (TFB) is a comprehensive and diverse dataset collection designed to facilitate the empirical evaluation and comparison of Time Series Forecasting (TSF) methods with enhanced fairness. It addresses limitations in existing benchmarks and offers a broad coverage of existing methods, spanning statistical learning, machine learning, and deep learning approaches. TFB aims to provide researchers with a robust and extensive evaluation platform, addressing dataset bias and limited coverage prevalent in existing benchmarks.