DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | 김우창 | - |
dc.contributor.author | Bae, Sanghyeon | - |
dc.contributor.author | 배상현 | - |
dc.date.accessioned | 2024-08-08T19:31:10Z | - |
dc.date.available | 2024-08-08T19:31:10Z | - |
dc.date.issued | 2024 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=1099248&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/322036 | - |
dc.description | 학위논문(박사) - 한국과학기술원 : 산업및시스템공학과, 2024.2,[iv, 84 p. :] | - |
dc.description.abstract | This dissertation studies backgrounds and methodologies to develop modeling and computational techniques for personalized financial decision making problems. First, we construct a scalable coupon personalization model for a credit card company. The main challenges include dealing with a large number of target stores and customers, as well as encoding the diverse customer and store data into meaningful latent features. Our machine-learning based model address these challenges, and is practically integrated into the real-world coupon targeting system for a credit card company in South Korea. Second, we analyze a portfolio choice problem for a couple with tax-deferred accounts (TDAs) and survival-contingent products (SCPs). Unlike previous research that focused on only one type of product, we explore both TDAs and SCPs, which are commonly used in practice. To address curse of dimensionality issue due to high-dimensional state space and long planning horizon, we apply the stochastic dual dynamic programming (SDDP) algorithm with modified sampling strategies. Lastly, we introduce a novel goal-based investing (GBI) model with goal postponement using a multistage mixed-integer stochastic program. Unlike conventional approaches, our model allows individuals the flexibility to defer goal fulfillment within a specific timeframe. Our study presents both theoretical and empirical findings associated with goal postponement in the context of GBI. We also introduce the stochastic dual dynamic integer programming (SDDiP) algorithm to solve practical large-scale GBI problem. We expect our studies will provide efficient computational methodologies and practical findings in complex and large-scale personalized financial decision making problems. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | 개인화▼a재무 의사결정▼a머신러닝▼a대규모 최적화▼a다기간 추계 계획법▼a분해 기법▼a목표 기반 투자 | - |
dc.subject | Personalization▼aFinancial decision making▼aMachine learning▼aLarge-scale optimization▼aMultistage stochastic program▼aDecomposition methods▼aGoal-based investing | - |
dc.title | Modeling and computational methods for personalized financial decision making: machine learning, optimization, and decomposition techniques | - |
dc.title.alternative | 개인화된 재무 의사결정을 위한 모델링과 계산적 방법론: 머신러닝, 최적화, 분해기법을 중심으로 | - |
dc.type | Thesis(Ph.D) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :산업및시스템공학과, | - |
dc.contributor.alternativeauthor | Kim, Woo Chang | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.