Bounded-Abstention Multi-horizon Time-series Forecasting

TL;DR

This paper introduces a novel framework for multi-horizon time-series forecasting with abstention, allowing models to abstain from predictions when uncertain, and provides algorithms that outperform existing methods across multiple datasets.

Contribution

It formalizes the learning with abstention problem for multi-horizon forecasting, proposing new abstention strategies tailored to the structured nature of the task.

Findings

Proposed algorithms outperform baselines on 24 datasets.

Three natural abstention notions are theoretically analyzed and implemented.

Structured abstention improves forecast reliability in high-stakes applications.

Abstract

Multi-horizon time-series forecasting involves simultaneously making predictions for a consecutive sequence of subsequent time steps. This task arises in many application domains, such as healthcare and finance, where mispredictions can have a high cost and reduce trust. The learning with abstention framework tackles these problems by allowing a model to abstain from offering a prediction when it is at an elevated risk of making a misprediction. Unfortunately, existing abstention strategies are ill-suited for the multi-horizon setting: they target problems where a model offers a single prediction for each instance. Hence, they ignore the structured and correlated nature of the predictions offered by a multi-horizon forecaster. We formalize the problem of learning with abstention for multi-horizon forecasting setting and show that its structured nature admits a richer set of abstention problems. Concretely, we propose three natural notions of how a model could abstain for multi-horizon forecasting. We theoretically analyze each problem to derive the optimal abstention strategy and propose an algorithm that implements it. Extensive evaluation on 24 datasets shows that our proposed algorithms significantly outperforms existing baselines.