English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In data-driven stochastic optimization, model parameters of the underlyingdistribution need to be estimated from data in addition to the optimizationtask. Recent literature considers integrating the estimation and optimizationprocesses by selecting model parameters that lead to the best empiricalobjective performance. This integrated approach, which we callintegrated-estimation-optimization (IEO), can be readily shown to outperformsimple estimate-then-optimize (ETO) when the model is misspecified. In thispaper, we show that a reverse behavior appears when the model class iswell-specified and there is sufficient data. Specifically, for a general classof nonlinear stochastic optimization problems, we show that simple ETOoutperforms IEO asymptotically when the model class covers the ground truth, inthe strong sense of stochastic dominance of the regret. Namely, the entiredistribution of the regret, not only its mean or other moments, is alwaysbetter for ETO compared to IEO. Our results also apply to constrained,contextual optimization problems where the decision depends on observedfeatures. Whenever applicable, we also demonstrate how standard sample averageapproximation (SAA) performs the worst when the model class is well-specifiedin terms of regret, and best when it is misspecified. Finally, we provideexperimental results to support our theoretical comparisons and illustrate whenour insights hold in finite-sample regimes and under various degrees ofmisspecification.

Estimate-Then-Optimize versus Integrated-Estimation-Optimization versus Sample Average Approximation: A Stochastic Dominance Perspective