Transparent Networks for Multivariate Time Series

TL;DR

This paper introduces GATSM, a transparent neural network model for multivariate time series that combines interpretability with strong predictive performance, addressing a key gap in high-stakes applications.

Contribution

The paper proposes GATSM, a novel transparent neural network architecture for time series that effectively captures temporal patterns while maintaining interpretability.

Findings

GATSM outperforms existing generalized additive models.

GATSM achieves comparable performance to black-box models like RNNs and Transformers.

GATSM uncovers meaningful patterns in time series data.

Abstract

Transparent models, which provide inherently interpretable predictions, are receiving significant attention in high-stakes domains. However, despite much real-world data being collected as time series, there is a lack of studies on transparent time series models. To address this gap, we propose a novel transparent neural network model for time series called Generalized Additive Time Series Model (GATSM). GATSM consists of two parts: 1) independent feature networks to learn feature representations, and 2) a transparent temporal module to learn temporal patterns across different time steps using the feature representations. This structure allows GATSM to effectively capture temporal patterns and handle varying-length time series while preserving transparency. Empirical experiments show that GATSM significantly outperforms existing generalized additive models and achieves comparable performance to black-box time series models, such as recurrent neural networks and Transformer. In addition, we demonstrate that GATSM finds interesting patterns in time series.