GenTS: A Comprehensive Benchmark Library for Generative Time Series Models

TL;DR

GenTS is an extensive, flexible benchmarking library tailored for evaluating generative models in time series analysis, addressing the limitations of existing tools designed mainly for discriminative models.

Contribution

The paper introduces GenTS, a modular, comprehensive benchmark library specifically designed for systematic assessment of generative time series models, with unified pipelines and evaluation metrics.

Findings

Benchmarking results provide insights into model performance across tasks.

GenTS facilitates model comparison and selection.

The library supports customization for diverse research needs.

Abstract

Generative models have demonstrated remarkable potential in time series analysis tasks, like synthesis, forecasting, imputation, etc. However, offering limited coverage for generative models, existing time series libraries are mainly engineered for discriminative models, with standardized workflows for specific tasks, such as optimizing Mean Squared Errors for time series forecasting. This rigid structure is fundamentally incompatible with the distinct and often complex paradigms of generative models (e.g., adversarial training, diffusion processes), which learn the underlying data distribution rather than a direct input-output mapping. To this end, we proposed GenTS, a comprehensive and extensible benchmark library designed for systematic assessment on generative time series models. GenTS features a unified data preprocessing pipeline, a collection of versatile models, and panoramic evaluation metrics. Its modular design also enables the researchers to flexibly customize beyond our built-in datasets and models. Based on GenTS, we conducted benchmarking experiments under diverse tasks, accordingly offering suggestions for model selection and identifying potential directions for future research. Our codes are open-source at https://github.com/WillWang1113/GenTS. The official tutorials and document are available at https://willwang1113.github.io/GenTS/.