Distributionally robust stochastic programs with side information based on trimmings
This is a research paper whose authors are Adrián Esteban-Pérez and Juan M. Morales.
Abstract:
- We look at stochastic programmes that are conditional on some covariate information, where the only knowledge of the possible relationship between the unknown parameters and the covariates is a limited data sample of their joint distribution. We build a data-driven Distributionally Robust Optimization (DRO) framework to hedge the decision against the inherent error in the process of inferring conditional information from limited joint data by leveraging the close relationship between the notion of trimmings of a probability measure and the partial mass transportation problem.
- We demonstrate that our technique is computationally as tractable as the usual (no side information) Wasserstein-metric-based DRO and provides performance guarantees. Furthermore, our DRO framework may be easily applied to data-driven decision-making issues involving tainted samples. Finally, using a single-item newsvendor problem and a portfolio allocation problem with side information, the theoretical findings are presented.
Conclusions:
- We used the relationship between probability reductions and partial mass transit in this study to give a straightforward, yet powerful and creative technique to expand the usual Wasserstein-metric-based DRO to the situation of conditional stochastic programming. In the process of inferring the conditional probability measure of the random parameters from a limited sample drawn from the genuine joint data-generating distribution, our technique generates judgments that are distributionally resilient to uncertainty. In a series of numerical tests based on the single-item newsvendor issue and a portfolio allocation problem, we proved that our strategy achieves much higher out-of-sample performance than several current options. We backed up these actual findings with theoretical analysis, demonstrating that our strategy had appealing performance guarantees.
To learn more about probability, visit :
brainly.com/question/11234923
#SPJ4
Answer:
for brand x for 1 fluid ounces
for brand y
0,75-0,65=0,1
Brand x costs $0,1 less
Step-by-step explanation:
313.75 I believe that is the correct answer or som
<h2>
Answer:</h2><h2>
The actual height of the building = 36 feet.</h2>
Step-by-step explanation:
By blue print conversion,
2 inches = 8 feet
Therefore, 1 inch =
= 4 feet
1 inch = 4 feet
The height of the building on the blueprint = 9 inches
The actual height of the building = 9 * 4 = 36 feet.
<h2>D. This is an experiment with blocking </h2><h2 />
For your question: "A teacher wants to investigate whether listening to music while taking a math test increases scores. He instructs half
of his 36 students to listen to music while taking a test and the other half to not listen to music during the test. To
determine who listens and who does not, the teacher identifies the two students who scored highest on the last test
and randomly instructs one to listen to music and one to not. The teacher does the same for the next two highest-
scoring students and continues in this manner until each student has their instructions. Which of the following best
describes this plan?"
The answer is D