PYRREGULAR: A Unified Framework for Irregular Time Series, with Classification Benchmarks

2025-05-09

AI Summary

Irregular time series—characterized by asynchronous sampling, pervasive missing values, and variable lengths—pose significant challenges in modeling across healthcare, transportation, and environmental monitoring, exacerbated by fragmented evaluation protocols and isolated method development. To address this, we introduce IrregTimeBench, the first unified benchmark framework for irregular time series analysis. It comprises: (1) a standardized repository of 34 real-world datasets spanning diverse domains and irregularity patterns; (2) a unified NumPy/PyTorch array interface with missingness-aware temporal encoding; and (3) a reproducible evaluation pipeline integrating 12 model families—including RNNs, Neural ODEs, and TS2Vec. Comprehensive experiments systematically characterize model robustness across distinct irregularity types, substantially improving method comparability, framework interoperability, and evaluation consistency. IrregTimeBench establishes a new standard for rigorous, transparent, and reproducible research in irregular time series modeling.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization

Abstract

Irregular temporal data, characterized by varying recording frequencies, differing observation durations, and missing values, presents significant challenges across fields like mobility, healthcare, and environmental science. Existing research communities often overlook or address these challenges in isolation, leading to fragmented tools and methods. To bridge this gap, we introduce a unified framework, and the first standardized dataset repository for irregular time series classification, built on a common array format to enhance interoperability. This repository comprises 34 datasets on which we benchmark 12 classifier models from diverse domains and communities. This work aims to centralize research efforts and enable a more robust evaluation of irregular temporal data analysis methods.

Problem

• Challenges in analyzing irregular temporal data with varying frequencies
• Lack of unified tools for irregular time series classification
• Need for standardized benchmarks to evaluate analysis methods

Innovation

• Unified framework for irregular time series
• Standardized dataset repository introduced
• Benchmarked 12 classifier models