Saved in:
Bibliographic Details
Main Authors: Heuton, Kyle, Muench, F. Samuel, Shrestha, Shikhar, Stopka, Thomas J., Hughes, Michael C.
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.05622
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918187141955584
author Heuton, Kyle
Muench, F. Samuel
Shrestha, Shikhar
Stopka, Thomas J.
Hughes, Michael C.
author_facet Heuton, Kyle
Muench, F. Samuel
Shrestha, Shikhar
Stopka, Thomas J.
Hughes, Michael C.
contents Optimal allocation of scarce resources is a common problem for decision makers faced with choosing a limited number of locations for intervention. Spatiotemporal prediction models could make such decisions data-driven. A recent performance metric called fraction of best possible reach (BPR) measures the impact of using a model's recommended size K subset of sites compared to the best possible top-K in hindsight. We tackle two open problems related to BPR. First, we explore how to rank all sites numerically given a probabilistic model that predicts event counts jointly across sites. Ranking via the per-site mean is suboptimal for BPR. Instead, we offer a better ranking for BPR backed by decision theory. Second, we explore how to train a probabilistic model's parameters to maximize BPR. Discrete selection of K sites implies all-zero parameter gradients which prevent standard gradient training. We overcome this barrier via advances in perturbed optimizers. We further suggest a training objective that combines likelihood with a decision-aware BPR constraint to deliver high-quality top-K rankings as well as good forecasts for all sites. We demonstrate our approach on two where-to-intervene applications: mitigating opioid-related fatal overdoses for public health and monitoring endangered wildlife.
format Preprint
id arxiv_https___arxiv_org_abs_2503_05622
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Decision-aware training of spatiotemporal forecasting models to select a top K subset of sites for intervention
Heuton, Kyle
Muench, F. Samuel
Shrestha, Shikhar
Stopka, Thomas J.
Hughes, Michael C.
Machine Learning
Optimal allocation of scarce resources is a common problem for decision makers faced with choosing a limited number of locations for intervention. Spatiotemporal prediction models could make such decisions data-driven. A recent performance metric called fraction of best possible reach (BPR) measures the impact of using a model's recommended size K subset of sites compared to the best possible top-K in hindsight. We tackle two open problems related to BPR. First, we explore how to rank all sites numerically given a probabilistic model that predicts event counts jointly across sites. Ranking via the per-site mean is suboptimal for BPR. Instead, we offer a better ranking for BPR backed by decision theory. Second, we explore how to train a probabilistic model's parameters to maximize BPR. Discrete selection of K sites implies all-zero parameter gradients which prevent standard gradient training. We overcome this barrier via advances in perturbed optimizers. We further suggest a training objective that combines likelihood with a decision-aware BPR constraint to deliver high-quality top-K rankings as well as good forecasts for all sites. We demonstrate our approach on two where-to-intervene applications: mitigating opioid-related fatal overdoses for public health and monitoring endangered wildlife.
title Decision-aware training of spatiotemporal forecasting models to select a top K subset of sites for intervention
topic Machine Learning
url https://arxiv.org/abs/2503.05622