Towards Accurate and Interpretable Time-series Forecasting: A Polynomial Learning Approach

2026-03-03

AI Summary

Existing time series forecasting methods struggle to simultaneously achieve high prediction accuracy and feature-level interpretability, limiting user trust and the effectiveness of early warning systems. To address this challenge, this work proposes Interpretable Polynomial Learning (IPL), a novel approach that explicitly models raw features and their arbitrary-order interactions through a polynomial representation embedded directly within the model architecture. This design inherently preserves temporal dependencies while endowing the model with intrinsic interpretability. By adjusting the polynomial order, IPL offers a flexible trade-off between predictive accuracy and interpretability. Experimental results on synthetic data, Bitcoin prices, and real-world antenna measurements demonstrate that IPL not only maintains high forecasting accuracy but also significantly outperforms existing interpretable methods, enabling the construction of more concise and effective early warning mechanisms.

Technology Category

• Machine Learning: Feature Construction/Reformulation
• Computer Vision: Interpretability, Explainability, and Transparency
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Time series forecasting enables early warning and has driven asset performance management from traditional planned maintenance to predictive maintenance. However, the lack of interpretability in forecasting methods undermines users' trust and complicates debugging for developers. Consequently, interpretable time-series forecasting has attracted increasing research attention. Nevertheless, existing methods suffer from several limitations, including insufficient modeling of temporal dependencies, lack of feature-level interpretability to support early warning, and difficulty in simultaneously achieving the accuracy and interpretability. This paper proposes the interpretable polynomial learning (IPL) method, which integrates interpretability into the model structure by explicitly modeling original features and their interactions of arbitrary order through polynomial representations. This design preserves temporal dependencies, provides feature-level interpretability, and offers a flexible trade-off between prediction accuracy and interpretability by adjusting the polynomial degree. We evaluate IPL on simulated and Bitcoin price data, showing that it achieves high prediction accuracy with superior interpretability compared with widely used explainability methods. Experiments on field-collected antenna data further demonstrate that IPL yields simpler and more efficient early warning mechanisms.

Problem

• time-series forecasting
• interpretability
• temporal dependencies
• feature-level interpretability
• accuracy-interpretability trade-off

Innovation

• interpretable time-series forecasting
• polynomial learning
• feature-level interpretability
• temporal dependencies
• predictive maintenance