FreqLens: Interpretable Frequency Attribution for Time Series Forecasting

2026-02-09

AI Summary

This work addresses the limited interpretability of time series forecasting models, which hinders their deployment in high-stakes applications requiring trustworthy predictions. To this end, the authors propose FreqLens, a framework that explicitly attributes predictions to dominant periodic components automatically discovered in the data. FreqLens integrates a learnable frequency discovery mechanism with a Shapley value-based attribution method that satisfies key axioms such as completeness and faithfulness. By employing Sigmoid-parameterized frequency bases and diversity regularization, the approach enables theoretically grounded, frequency-level knowledge discovery. Experiments on Traffic and Weather datasets demonstrate that FreqLens achieves strong predictive performance while accurately identifying physically meaningful periodicities—such as 24-hour, 12-hour, and weekly cycles—thereby offering both accuracy and interpretability.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Computer Vision: Interpretability, Explainability, and Transparency
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Time series forecasting models often lack interpretability, limiting their adoption in domains requiring explainable predictions. We propose \textsc{FreqLens}, an interpretable forecasting framework that discovers and attributes predictions to learnable frequency components. \textsc{FreqLens} introduces two key innovations: (1) \emph{learnable frequency discovery} -- frequency bases are parameterized via sigmoid mapping and learned from data with diversity regularization, enabling automatic discovery of dominant periodic patterns without domain knowledge; and (2) \emph{axiomatic frequency attribution} -- a theoretically grounded framework that provably satisfies Completeness, Faithfulness, Null-Frequency, and Symmetry axioms, with per-frequency attributions equivalent to Shapley values. On Traffic and Weather datasets, \textsc{FreqLens} achieves competitive or superior performance while discovering physically meaningful frequencies: all 5 independent runs discover the 24-hour daily cycle ($24.6 \pm 0.1$h, 2.5\% error) and 12-hour half-daily cycle ($11.8 \pm 0.1$h, 1.6\% error) on Traffic, and weekly cycles ($10\times$ longer than the input window) on Weather. These results demonstrate genuine frequency-level knowledge discovery with formal theoretical guarantees on attribution quality.

Problem

• time series forecasting
• interpretability
• frequency attribution
• explainable AI

Innovation

• learnable frequency discovery
• axiomatic frequency attribution
• time series forecasting
• Shapley values
• interpretable AI