# Learning Decisions Offline from Censored Observations with ε-insensitive Operational Costs

arXiv: https://arxiv.org/abs/2408.07305 · HTML: https://mlsanity.com/paper/2408.07305/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 52 references that are in the map.
- First posted on arXiv 14 Aug 2024.

## Builds on (most cited first, 9 of 9)

- [Optimization Methods for Large-Scale Machine Learning](https://mlsanity.com/paper/1606.04838/) (2016)
- [Train faster, generalize better: Stability of stochastic gradient descent](https://mlsanity.com/paper/1509.01240/) (2015)
- [Stability of Stochastic Gradient Descent on Nonsmooth Convex Losses](https://mlsanity.com/paper/2006.06914/) (2020)
- [From Predictive to Prescriptive Analytics](https://mlsanity.com/paper/1402.5481/) (2014)
- [High probability generalization bounds for uniformly stable algorithms with nearly optimal rate](https://mlsanity.com/paper/1902.10710/) (2019)
- [JANOS: An Integrated Predictive and Prescriptive Modeling Framework](https://mlsanity.com/paper/1911.09461/) (2019)
- [Applying Deep Learning to the Newsvendor Problem](https://mlsanity.com/paper/1607.02177/) (2016)
- [Risk Guarantees for End-to-End Prediction and Optimization Processes](https://mlsanity.com/paper/2012.15046/) (2020)
- [Predictive and Prescriptive Analytics for Location Selection of Add-on Retail Products](https://mlsanity.com/paper/1804.01182/) (2018)

## Abstract (arXiv)

Many important managerial decisions are made based on censored observations. Making decisions without adequately handling the censoring leads to inferior outcomes. We investigate the data-driven decision-making problem with an offline dataset containing the feature data and the censored historical data of the variable of interest without the censoring indicators. Without assuming the underlying distribution, we design and leverage ε-insensitive operational costs to deal with the unobserved censoring in an offline data-driven fashion. We demonstrate the customization of the ε-insensitive operational costs for a newsvendor problem and use such costs to train two representative ML models, including linear regression (LR) models and neural networks (NNs). We derive tight generalization bounds for the custom LR model without regularization (LR-εNVC) and with regularization (LR-εNVC-R), and a high-probability generalization bound for the custom NN (NN-εNVC) trained by stochastic gradient descent. The theoretical results reveal the stability and learnability of LR-εNVC, LR-εNVC-R and NN-εNVC. We conduct extensive numerical experiments to compare LR-εNVC-R and NN-εNVC with two existing approaches, estimate-as-solution (EAS) and integrated estimation and optimization (IEO). The results show that LR-εNVC-R and NN-εNVC outperform both EAS and IEO, with maximum cost savings up to 14.40% and 12.21% compared to the lowest cost generated by the two existing approaches. In addition, LR-εNVC-R's and NN-εNVC's order quantities are statistically significantly closer to the optimal solutions should the underlying distribution be known.