[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-121776-en":3,"doc-seo-121776-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},121776,8796095360427,"Lucas Martin","https://ap-avatar.wpscdn.com/davatar_994ba38a5ba835b3df7d355c54d3ed8d",8,"Research & Report","Nonparametric Discrete Choice Experiments with Machine Learning Guided Adaptive Design","Designing products that match consumers’ preferences is crucial for business success. The work introduces Gradient-based Survey (GBS), a discrete choice experiment for multi-attribute product design that gathers preferences via adaptive sequences of paired comparisons on partial profiles. GBS avoids the traditional parametric random-utility model, improving robustness to model misspecification. By combining machine learning gradients with DCEs, it scales to hundreds of attributes and supports personalized design for heterogeneous consumers, demonstrating gains in inaccuracy and sample efficiency in simulations.","arXiv :2310 . 12026v1 [ stat .ML] 18 Oct 2023  \nNonparametric Discrete Choice Experiments with Machine Learning Guided Adaptive Design  \nMingzhang Yin1 , Ruijiang Gao2 , Weiran Lin1 , and Steven M. Shugan1  \n1 University of Florida, Warrington College of Business  \n2 University of Texas at Austin, McCombs School of Business  \nOctober 19, 2023  \nAbstract  \nDesigning products to meet consumers’ preferences is essential for a business’s success. We propose Gradient-based Survey (GBS), a discrete choice experiment for multiattribute product design. The experiment elicits consumer preferences through a sequence of paired comparisons for partial profiles. GBS adaptively constructs paired comparison questions based on the respondents’ previous choices. Unlike the traditional random utility maximization paradigm, GBS is robust to model misspecification by not requiring a parametric utility model. Cross-pollinating the machine learning and experiment design, GBS is scalable to products with hundreds of attributes and can design personalized products for heterogeneous consumers. We demonstrate the advantage of GBS inaccuracy and sample efficiency compared to the existing parametric and nonparametric methods in simulations.  \n1 Introduction  \nIdentifying an optimal product based on consumers’ preferences is essential for the success of a business. Such a problem is prevalent in companies where the product consists of multiple attributes such as health insurance, cell phone plans, pizzas, automobiles, logos, and email advertisements (Balakrishnan et al., 2004; Bertsimas and Miši, 2017; Ellicksonet al., 2022; Netzer and Srinivasan, 2011) . In this paper, we focus on products with discrete attributes represented as multivariate binary variables, as is the case in A/B testing 1. Selecting appropriate products from a choice set is complicated for decision-makers. First, the desired product should meet the unobserved consumer preferences. The latent preferences are often revealed by survey experiments known as conjoint analysis (Green and Rao, 1971; Luce and Tukey, 1964) . Since the milestone work of Louviere and Woodworth (1983), the choice-based conjoint (CBC) analysis become one of the most widely used methods to quantify multiattribute preference (Hein et al., 2020) . A prevalent assumption is that the preference for an attribute is quantified by a part-worth score, and the total utility of a product profile is the sum of part-worths (Green and Srinivasan, 1978) . This parametric assumption, however, may oversimplify how respondents encode and evaluate products (Allenby et al., 2005) . Second, the choice set for product design grows exponentially with the number of attributes, making the problem NP-hard (Kohli and Krishnamurti, 1989) . The problem is more evident with the development of technology when more and more components are integrated into a single product. For example, the design of a smartphone might need to consider hundreds of attributes from a digital camera, screen display, connectivity modules, software applications, and a range of sensors. The high-dimensional attributes  \n1A discrete attribute with more than two levels can be coded as multiple binary variables.  \nFigure 1: Demonstration of a paired choice question for a logo design.  \npose a scalability challenge to the extant product design methods. Lastly, consumer preference is heterogeneous, and it is desired to provide a customized product aligned with individual tastes. Nowadays, an increasing number of products are presented as digital content, making personalized product design feasible. For example, the email campaign content can be designed based on a receiver’s demographic information. The fast development of generative AI might expand the need for personalized product design.  \nWe propose GBS that is robust to model misspecification of the utility function, scalable to high-dimensional attributes and applies to single or personalized product design. The idea of ","cbCaiv6IP9xgBo6B","https://ap.wps.com/l/cbCaiv6IP9xgBo6B","pdf",593663,1,13,"English","en",105,"# Introduction\n## Problem setting in multi-attribute product choice\n## Limitations of parametric utility models\n## Scalability and preference heterogeneity\n# Proposed approach: Gradient-based Survey (GBS)\n## Gradient estimation via score function\n## Adaptive question construction with SGD\n## Variance reduction and experiment efficiency","[{\"question\":\"What is Gradient-based Survey (GBS) in discrete choice experiments?\",\"answer\":\"GBS is an adaptive discrete choice experiment that elicits preferences through sequential paired comparisons on partial profiles. It constructs the next questions based on respondents’ previous choices.\"},{\"question\":\"How does GBS differ from traditional parametric random utility models?\",\"answer\":\"GBS is robust to model misspecification because it does not require a parametric utility model. It instead computes gradients using a score-function approach from DCE data.\"},{\"question\":\"What scalability and personalization capabilities does the proposed method offer?\",\"answer\":\"GBS scales to products with hundreds of attributes and can generate personalized products for heterogeneous consumers. Simulations show improved inaccuracy and sample efficiency compared with existing methods.\"}]","Nonparametric Discrete Choice Experiments with Machine Learning Guided Adaptive Design | PDF",1785806779,33,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"nonparametric-discrete-choice-experiments-with-machine-learning-guided-adaptive-design","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/nonparametric-discrete-choice-experiments-with-machine-learning-guided-adaptive-design/121776/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-04",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What is Gradient-based Survey (GBS) in discrete choice experiments?","Question",{"text":75,"@type":76},"GBS is an adaptive discrete choice experiment that elicits preferences through sequential paired comparisons on partial profiles. It constructs the next questions based on respondents’ previous choices.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does GBS differ from traditional parametric random utility models?",{"text":80,"@type":76},"GBS is robust to model misspecification because it does not require a parametric utility model. It instead computes gradients using a score-function approach from DCE data.",{"name":82,"@type":73,"acceptedAnswer":83},"What scalability and personalization capabilities does the proposed method offer?",{"text":84,"@type":76},"GBS scales to products with hundreds of attributes and can generate personalized products for heterogeneous consumers. Simulations show improved inaccuracy and sample efficiency compared with existing methods.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]