Dynamic Combination of Heterogeneous Models for Hierarchical Time Series

TL;DR

This paper presents DYCHEM, a framework that dynamically combines diverse models for hierarchical time series forecasting, improving accuracy, coherence, and reliability across various datasets.

Contribution

DYCHEM introduces a novel method for integrating heterogeneous models and hierarchical structure learning, enhancing forecast accuracy and coherence in hierarchical time series.

Findings

DYCHEM outperforms existing methods on public and financial datasets.

The framework produces coherent quantile forecasts independent of model choice.

It demonstrates robustness and efficiency in large-scale forecasting tasks.

Abstract

We introduce a framework to dynamically combine heterogeneous models called \texttt{DYCHEM}, which forecasts a set of time series that are related through an aggregation hierarchy. Different types of forecasting models can be employed as individual ``experts'' so that each model is tailored to the nature of the corresponding time series. \texttt{DYCHEM} learns hierarchical structures during the training stage to help generalize better across all the time series being modeled and also mitigates coherency issues that arise due to constraints imposed by the hierarchy. To improve the reliability of forecasts, we construct quantile estimations based on the point forecasts obtained from combined heterogeneous models. The resulting quantile forecasts are coherent and independent of the choice of forecasting models. We conduct a comprehensive evaluation of both point and quantile forecasts for hierarchical time series (HTS), including public data and user records from a large financial software company. In general, our method is robust, adaptive to datasets with different properties, and highly configurable and efficient for large-scale forecasting pipelines.