TSPP: A Unified Benchmarking Tool for Time-series Forecasting

TL;DR

This paper introduces TSPP, a comprehensive benchmarking framework for time series forecasting that standardizes evaluation and facilitates comparison of diverse models, including deep learning and decision trees.

Contribution

The paper presents a unified benchmarking tool for time series forecasting that integrates models and datasets, enabling fair comparison and assessment of different approaches.

Findings

Deep learning models can match gradient-boosting decision trees with minimal effort.

The framework promotes fair and standardized evaluation of forecasting methods.

Benchmarking reveals the competitive performance of recent models.

Abstract

While machine learning has witnessed significant advancements, the emphasis has largely been on data acquisition and model creation. However, achieving a comprehensive assessment of machine learning solutions in real-world settings necessitates standardization throughout the entire pipeline. This need is particularly acute in time series forecasting, where diverse settings impede meaningful comparisons between various methods. To bridge this gap, we propose a unified benchmarking framework that exposes the crucial modelling and machine learning decisions involved in developing time series forecasting models. This framework fosters seamless integration of models and datasets, aiding both practitioners and researchers in their development efforts. We benchmark recently proposed models within this framework, demonstrating that carefully implemented deep learning models with minimal effort can rival gradient-boosting decision trees requiring extensive feature engineering and expert knowledge.