GenTS: A Comprehensive Benchmark Library for Generative Time Series Models

Unknown Date

AI Summary

Existing time series libraries are primarily designed for discriminative models and struggle to accommodate the complex training paradigms—such as adversarial training and diffusion processes—and evaluation requirements of generative models. To address this gap, this work proposes the first modular, task-agnostic unified benchmarking framework for generative time series modeling. The framework integrates standardized data preprocessing pipelines, a diverse suite of generative architectures—including GANs and diffusion models—and multidimensional evaluation metrics, enabling flexible extension and multi-task assessment. Leveraging this framework, we conduct systematic experiments to offer practical model selection guidance, identify key challenges in generative time series modeling, and release open-source code with tutorials to foster standardization and reproducibility in the field.

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/.

Problem

• generative time series models
• benchmarking
• time series analysis
• model evaluation
• generative modeling

Innovation

• Generative Time Series Models
• Benchmark Library
• Modular Design
• Unified Evaluation
• Diffusion Processes