Fast Rates for Contextual Linear Optimization

Incorporating side observations in decision making can reduce uncertainty and boost performance, but it also requires that we tackle a potentially complex predictive relationship. Although one may use off-the-shelf machine learning methods to separately learn model plug in, variety of recent instead integrate estimation optimization by fitting the directly optimize downstream performance. Surprisingly, case contextual linear optimization, show naïve plug-in approach actually achieves regret convergence rates are significantly faster than We this leveraging fact specific problem instances do not have arbitrarily bad near-dual-degeneracy. there other pros cons consider as discuss illustrate numerically, our results highlight nuanced landscape for enterprise optimization. Our overall positive practice: models easy fast train using existing tools; simple interpret; and, show, lead decisions perform very well. This paper was accepted Hamid Nazerzadeh, data science.