CATF: A Manager-Worker Framework for Context-Aware Time-Series Forecasting

Forum topic · 小凯 · 2026-11-04

Summary

CATF is a Context-Aware Time-Series Forecasting framework built on a Manager-Worker architecture, proposed to handle time-series prediction where a single target variable depends on its own history plus contextual variables such as weather and calendar information. Unlike ensemble learning and Mixture-of-Experts approaches that rely on manual data partitioning or soft routing—which increases inference cost and hurts interpretability—CATF uses a Transformer-based Manager that hard-routes each input to a single specialized Worker during both training and inference. Workers therefore specialize without manual partitioning, and any single-model forecaster can serve as a Worker. Experiments across five context-rich datasets, four forecasting horizons, and two normalization schemes show CATF beating its standalone counterpart in 162 of 200 comparisons; the best CATF-enhanced model outperforms the best standalone baseline in 19 of 20 settings and the zero-shot Time-MoE foundation model in 15 of 20 settings. A bike-sharing case study shows the Manager routes by calendar while each Worker specializes in a distinct weekly pattern.

Paper Overview

Field: Machine Learning Authors: Yueqiao Chen, Dongyu Liu Published: 2026-11-03 arXiv: 2611.05325

Introduction

We study context-aware time-series forecasting, in which a single target variable is predicted from its own history and a set of contextual variables, such as weather and calendar information. As the context becomes more diverse, a single model struggles to capture all of its regimes, while ensemble learning and Mixture-of-Experts approaches rely on manual data partitioning or soft routing, which increases the inference cost and hinders interpretability.

The CATF Framework

We propose CATF, a Context-Aware Time-Series Forecasting framework built on a Manager-Worker architecture:

A Transformer-based Manager routes each input to a single Worker based on the context, using hard routing during both training and inference.
The Workers thus specialize without manual data partitioning.
Any single-model forecaster can serve as a Worker.

Results

Evaluated on five context-rich datasets, four horizons, and two normalization schemes.
CATF improves over its standalone counterpart in 162 of 200 comparisons.
The best CATF-enhanced model outperforms the best standalone baseline in 19 of 20 settings.
It outperforms the zero-shot Time-MoE foundation model in 15 of 20 settings.
A case study on bike-sharing data shows that the Manager routes inputs by the calendar and that each Worker specializes in a different weekly pattern.

Original Abstract

> We study context-aware time-series forecasting, in which a single target variable is predicted from its own history and a set of contextual variables, such as weather and calendar information. As the context becomes more diverse, a single model struggles to capture all of its regimes, while ensemble learning and Mixture-of-Experts approaches rely on manual data partitioning or soft routing, which increases the inference cost and hinders interpretability. We propose CATF, a Context-Aware Time-Series Forecasting framework built on a Manager-Worker architecture: a Transformer-based Manager routes each input to a single Worker based on the context, using hard routing during both training and inference. The Workers thus specialize without manual data partitioning, and any single-model forecaster can serve as a Worker. On five context-rich datasets, four horizons, and two normalization schemes, CATF improves over its standalone counterpart in 162 of 200 comparisons, the best CATF-enhanced model outperforms the best standalone baseline in 19 of 20 settings, and it outperforms the zero-shot Time-MoE foundation model in 15 of 20 settings. A case study on bike-sharing data shows that the Manager routes inputs by the calendar and that each Worker specializes in a different weekly pattern.

--- *Auto-collected on 2026-11-05*