[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117311-en":3,"doc-seo-117311-105":29,"detail-sidebar-cat-0-en-105":90},{"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":13,"seo_description":14,"update_tm":27,"read_time":28},117311,8796095461610,"Oliver","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Uncertainty-aware Machine Learning with Applications to Credit Risk - dissertation","Uncertainty-aware machine learning techniques are developed and assessed for credit risk modeling, with focus on loss given default (LGD) and market-based loss rates. The work quantifies predictive uncertainty by separating aleatoric and epistemic components, designs learning strategies for LGD prediction, and uses uncertainty-driven analysis to explain model outputs. Additional contributions address non-linearity and distributional behavior through feature selection, scenario analysis, robust modeling, and GAMME-Advances in predictive mean matching combined with simulation-based evaluation.","Uncertainty-aware Machine Learning with Applications to Credit Risk  \nA dissertation in partial fulfillment of the requirements for the degree of Doktor der Wirtschaftswissenschaft (Dr. rer. pol.)  \nsubmitted to the  \nFaculty of Business, Economics,  \nand Management Information Systems  \nUniversity Regensburg  \nsubmitted by  \nMatthias Nagl, M.Sc. in Survey Statistics  \nAdvisors  \nProf. Dr. Daniel Rösch Date of disputation  \nProf. Dr. Ralf Kellner November 4th, 2024  \nAcknowledgments  \nFirst and foremost, I would like to thank Prof. Dr. Daniel Rösch for his never ending support and encouragement to always improve my work. It was an honor to work with you on these projects. Thank you for taking the time to listen to my rough ideas and advising me what is really important to make a good paper out of them. Further, I want to express my gratitude to Prof. Dr. Ralf Kellner for being my second advisor. Thank you for your ongoing support and motivation.  \nI'm especially grateful for my colleagues at the University of Regensburg. You made this journey a wonderful and fascinating experience. Thank you all for always having an open door and answering my questions when I needed your help. Working on ideas is only really fun when you are doing it together.  \nFurther, I want to thank my brother and colleague Maximilian who served as a mentor in thescientific world to me. Thank you for your guidance and your continuous feedback. I get asked a lot: “How is it to work with your brother?” and I can say it works very well.  \nFurthermore, I want to thank my dear friends, Florian Scholze and Alexander Stemke, who accompanied me through my master and this journey. Doing a PhD can be challenging and it's good to have friends who can relate to it.  \nI dedicate this thesis to my amazing wife Regina. I'm most grateful for you and all the support you gave me. Thank you for raising me up every time I fell and listen to my thoughts about statistical problems that are only my problems. And lastly, I want to thank my parents, Lorenz and Kathrin, for letting me find my own way. The name by which we are known is part of our identity. And with this work, my dear father, I try to ensure that our name will never be forgotten.  \nContents  \nList of Figures iii  \nList of Tables vi  \nIntroduction 1  \n1 Quantifying uncertainty of machine learning methods for loss given default 10  \n1.1 Introduction ....................................... 11  \n1.2 Data ............................................ 14  \n1.3 Methods ......................................... 17  \n1.4 Results .......................................... 21  \n1.4.1 Learning strategy ................................ 21  \n1.4.2 Aleatoric and epistemic uncertainty in predictions ............. 24  \n1.4.3 Explaining LGD predictions .......................... 26  \n1.5 Conclusion ........................................ 28  \n2 Non-linearity and the distribution of market-based loss rates 30  \n2.1 Introduction ....................................... 31  \n2.2 Literature review .................................... 33  \n2.3 Data ............................................ 35  \n2.4 Methods ......................................... 37  \n2.5 Results .......................................... 44  \n2.5.1 Feature selection & model estimation ..................... 44  \n2.5.2 Drivers of 􀀖 and 􀀞 ................................ 48  \n2.5.3 Scenario analysis ................................ 54  \n2.6 Conclusion ........................................ 58  \n2.A Descriptive statistics .................................. 60  \n2.B Trainable activation functions ............................. 63  \n2.C Robustness ........................................ 65  \n2.D Scenario analysis ..................................... 68  \n3 GAMME-Advances in Predictive Mean Matching 70  \n3.1 Introduction ....................................... 71  \n3.2 Background ....................................... 72  \n3.2.1 Missing data mechanisms .....................","cbCaiiMbuluDfzdO","https://ap.wps.com/l/cbCaiiMbuluDfzdO","pdf",6001490,1,137,"English","en",105,"# Introduction\n## Quantifying uncertainty of machine learning methods for loss given default\n## Non-linearity and the distribution of market-based loss rates\n## GAMME-Advances in Predictive Mean Matching\n# Conclusion\n# References","[{\"question\":\"What is the main credit-risk target addressed in the dissertation?\",\"answer\":\"The dissertation focuses on modeling loss given default (LGD) and related market-based loss rates, aiming to improve prediction quality and interpretability under uncertainty.\"},{\"question\":\"How does the work treat uncertainty in machine learning predictions?\",\"answer\":\"It explicitly separates aleatoric and epistemic uncertainty in predictions and uses uncertainty estimation results to support further analysis and explanation of LGD outputs.\"},{\"question\":\"What methods are introduced for handling non-linearity and missing data?\",\"answer\":\"For non-linearity and market-based drivers, the dissertation uses feature selection and scenario analysis with dedicated robustness considerations. For missing data, it discusses missing-data mechanisms and introduces a methodology built around predictive mean matching, neural networks, and accumulated local effect plots.\"}]",1785675111,345,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"uncertainty-aware-machine-learning-with-applications-to-credit-risk-dissertation","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/uncertainty-aware-machine-learning-with-applications-to-credit-risk-dissertation/117311/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What is the main credit-risk target addressed in the dissertation?","Question",{"text":74,"@type":75},"The dissertation focuses on modeling loss given default (LGD) and related market-based loss rates, aiming to improve prediction quality and interpretability under uncertainty.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does the work treat uncertainty in machine learning predictions?",{"text":79,"@type":75},"It explicitly separates aleatoric and epistemic uncertainty in predictions and uses uncertainty estimation results to support further analysis and explanation of LGD outputs.",{"name":81,"@type":72,"acceptedAnswer":82},"What methods are introduced for handling non-linearity and missing data?",{"text":83,"@type":75},"For non-linearity and market-based drivers, the dissertation uses feature selection and scenario analysis with dedicated robustness considerations. For missing data, it discusses missing-data mechanisms and introduces a methodology built around predictive mean matching, neural networks, and accumulated local effect plots.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":105,"slug":137},19,"General","general"]