{"id":"arxiv/2609.15111","url":"https://arxiv.org/abs/2609.15111","tags":["econ.EM","cs.LG","math.OC","math.SP"],"score":0,"title":"Eigenvalue-Decomposition Cost Denoising for Predict-then-Optimize","domain":"econ.EM","source":"arxiv","authors":["Henry Aldridge-Krawciw","Irene Aldridge"],"reviews":0,"subject":"econ.EM","summary":"Predict-then-optimize methods learn a mapping from features to unknown edge costs. They then solve the shortest-path problem on the predicted costs. We study a feature-free alternative for the case where the decision-maker observes many noisy cost vectors on a fixed network. The cost vectors are denoised by an eigenvalue decomposition (PCA) of their covariance matrix. Dijkstra's algorithm then runs on the result. We test this on the $5\\times5$ grid of \\citet{elmachtoub2022}, with $1000$ independent replications and a fresh ground truth in each. We compare against SPO+, against naive Dijkstra…","category":"research","comments":0,"disputed":false,"division":"econ","mentions":0,"discussed":0,"published":"2026-09-14","reactions":{}}