Data-Driven Models for Uncertainty and Behavior
نویسندگان
چکیده
The last decade has seen an explosion in the availability of data. In this thesis, we propose new techniques to leverage these data to tractably model uncertainty and behavior. Specifically, this thesis consists of three parts: In the first part, we propose a novel schema for utilizing data to design uncertainty sets for robust optimization using hypothesis testing. The approach is flexible and widely applicable, and robust optimization problems built from our new datadriven sets are computationally tractable, both theoretically and practically. Optimal solutions to these problems enjoy a strong, finite-sample probabilistic guarantee. Computational evidence from classical applications of robust optimization – queuing and portfolio management – confirm that our new data-driven sets significantly outperform traditional robust optimization techniques whenever data is available. In the second part, we examine in detail an application of the above technique to the unit commitment problem. Unit commitment is a large-scale, multistage optimization problem under uncertainty that is critical to power system operations. Using real data from the New England market, we illustrate how our proposed data-driven uncertainty sets can be used to build high-fidelity models of the demand for electricity, and that the resulting large-scale, mixed-integer adaptive optimization problems can be solved efficiently. With respect to this second contribution, we propose new data-driven solution techniques for this class of problems inspired by ideas from machine learning. Extensive historical back-testing confirms that our proposed approach generates high quality solutions that compare with state-of-the-art methods. In the third part, we focus on behavioral modeling. Utility maximization (singleagent case) and equilibrium modeling (multi-agent case) are by far the most common behavioral models in operations research. By combining ideas from inverse optimization with the theory of variational inequalities, we develop an efficient, data-driven technique for estimating the primitives of these models. Our approach supports both parametric and nonparametric estimation through kernel learning. We prove that our estimators enjoy a strong generalization guarantee even when the model is misspecified. Finally, we present computational evidence from applications in economics and transportation science illustrating the effectiveness of our approach and its scalability
منابع مشابه
Event-driven and Attribute-driven Robustness
Over five decades have passed since the first wave of robust optimization studies conducted by Soyster and Falk. It is outstanding that real-life applications of robust optimization are still swept aside; there is much more potential for investigating the exact nature of uncertainties to obtain intelligent robust models. For this purpose, in this study, we investigate a more refined description...
متن کاملA model for illustrating the effective factors of strategic behavior (Case study: Social insurance company of Tehran)
Achieving the organizational objectives needs employees’ behavior and their collaboration with management. Strategic behavior-driven is depend on different factors. The aim of this article is to determine the impact of factors on strategic behavior-driven. A conceptual model was developed and tested on a survey in Social Insurance Company of Tehran which employees’ participated. Data collected ...
متن کاملAdaptive RBF network control for robot manipulators
TThe uncertainty estimation and compensation are challenging problems for the robust control of robot manipulators which are complex systems. This paper presents a novel decentralized model-free robust controller for electrically driven robot manipulators. As a novelty, the proposed controller employs a simple Gaussian Radial-Basis-Function Network as an uncertainty estimator. The proposed netw...
متن کاملA New Group Data Envelopment Analysis Method for Ranking Design Requirements in Quality Function Deployment
Data envelopment analysis (DEA) is an objective method for priority determination of decision making units (DMUs) with the same multiple inputs and outputs. DEA is an efficiency estimation technique, but it can be used for solving many problems of management such as rankig of DMUs. Many researchers have found similarity between DEA and MCDM techniques. One of the earliest techniques in MCDM is...
متن کاملComparison and evaluation of the performance of data-driven models for estimating suspended sediment downstream of Doroodzan Dam
Dams control most of the sediment entering the reservoir by creating static environments. However, sediment leaving the dam depends on various factors such as dam management method, inlet sediment, water height in the reservoir, the shape of the reservoir, and discharge flow. In this research, the amount of suspended sediment of Doroodzan Dam based on a statistical period of 25 years has been i...
متن کاملDiscrete time robust control of robot manipulators in the task space using adaptive fuzzy estimator
This paper presents a discrete-time robust control for electrically driven robot manipulators in the task space. A novel discrete-time model-free control law is proposed by employing an adaptive fuzzy estimator for the compensation of the uncertainty including model uncertainty, external disturbances and discretization error. Parameters of the fuzzy estimator are adapted to minimize the estimat...
متن کامل