Decision Sciences Theses and Dissertations

Permanent URI for this collection

https://hdl.handle.net/1794/12484

Browse

Browsing Decision Sciences Theses and Dissertations by Subject "Group Decision Making"

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    ESSAYS ON SUSTAINABLE SUPPLY CHAIN, GROUP DECISION MAKING AND EXPERT-AUGMENTED FEATURE SELECTION
    (University of Oregon, 2022-10-04) Rabiee, Meysam; Pangburn, Michael
    My dissertation consists of an essay in sustainable supply chain management, an essay in group decision making, and an essay in expert-augmented feature selection. My first essay is an unpublished work co-authored with Prof. Nagesh Murthy and Dr. Hossein Rikhtehgar Berenji. In some supply chains, it is extraordinarily expensive for a buyer to audit all selected suppliers to guarantee compliance with the buyer's code of conduct for social and environmental responsibility. In this work, we provide insight to help such a buyer profit from judicious audits, despite the risk of revenue loss due to non-compliance. My second essay is co-authored with Babak Aslani and Dr. Jafar Rezaei; it has been published. The purpose of this article is to investigate the detection and handling of biased decision-makers in group decision-making processes. To address this issue, we developed three algorithms including extreme, moderate, and soft versions. The third essay is an unpublished work co-authored with Mohsen Mirhashemi, Prof. Michael Pangburn, Prof. Dursun Delen, and Dr. Saeed Piri. In this study, we enrich the conventional feature-selection method by incorporating the opinions of experts on the features, a technique we refer to as expert-augmented feature selection. To reflect the trade-off between explainability and prediction accuracy, we develop two very similar models: one for classification problems, and one for regression problems. Finally, we develop a posterior ensemble approach for quantifying each feature's accuracy-contribution degree. This algorithm's output helps us to discover features that are consistently rated, under-rated, or over-rated by experts.