Mixing It Up: Exploring Mixer Networks for Irregular Multivariate Time Series Forecasting

TL;DR

The paper addresses forecasting irregularly sampled multivariate time series by introducing IMTS-Mixer, an all-MLP architecture that converts irregular channel observations into fixed-size embeddings via ISCAM and then forecasts at arbitrary times using Continuous Temporal Projection (ConTP). The approach achieves state-of-the-art accuracy on both PMAU and Physiome-ODE benchmarks while improving computational efficiency and reducing parameter count relative to competing methods. Through comprehensive ablations, the work demonstrates the effectiveness of the ISCAM encoder and the ConTP decoder, and analyzes the impact of mixer depth and component choices. These results highlight a practical, scalable alternative for IMTS forecasting with broad potential applications in healthcare, climate science, and biology.

Abstract

Forecasting Irregular Multivariate Time Series (IMTS) has recently emerged as a distinct research field, necessitating specialized models to address its unique challenges. While most forecasting literature assumes regularly spaced observations without missing values, many real-world datasets - particularly in healthcare, climate research, and biomechanics - violate these assumptions. Time Series (TS)-mixer models have achieved remarkable success in regular multivariate time series forecasting. However, they remain unexplored for IMTS due to their requirement for complete and evenly spaced observations. To bridge this gap, we introduce IMTS-Mixer, a novel forecasting architecture designed specifically for IMTS. Our approach retains the core principles of TS mixer models while introducing innovative methods to transform IMTS into fixed-size matrix representations, enabling their seamless integration with mixer modules. We evaluate IMTS-Mixer on a benchmark of four real-world datasets from various domains. Our results demonstrate that IMTS-Mixer establishes a new state-of-the-art in forecasting accuracy while also improving computational efficiency.

Figures (3)

• Figure 1: Example of an IMTS Forecasting task. The observations and forecasting targets are irregularly spaced.
• Figure 2: Overview of IMTS-Mixer's architecture
• Figure 3: Test MSE with varying number of mixer blocks.