Publications

Detailed Information

Vehicle Fuel-type Preference Related Elements, Comparing their Importance Depending on Each Fuel-type : 자동차 연료유형 선호도와 관련된 요인들의 연료유형별 중요도 차이 비교

DC Field Value Language
dc.contributor.advisor구윤모-
dc.contributor.author김민아-
dc.date.accessioned2023-06-29T02:01:58Z-
dc.date.available2023-06-29T02:01:58Z-
dc.date.issued2023-
dc.identifier.other000000175496-
dc.identifier.urihttps://hdl.handle.net/10371/193403-
dc.identifier.urihttps://dcollection.snu.ac.kr/common/orgView/000000175496ko_KR
dc.description학위논문(석사) -- 서울대학교대학원 : 공과대학 협동과정 기술경영·경제·정책전공, 2023. 2. 구윤모.-
dc.description.abstractThis study aims to present a new method for analyzing conjoint survey data to understand the differences in preference-related elements (PREs) depending on four major types of vehicle fuel: gasoline, diesel, electric, and hydrogen. PREs include vehicle features, such as price and class, and customer features, such as age and gender. Previous studies have been focused on discovering the list of PREs. However, comparing the importance of each PRE on preference toward certain vehicle fuel-type has yet to be sufficiently studied. Understanding the relative importance of PREs is required to make effective strategies and policies, especially for efficient target segmentation. In addition, this study included vehicle status quo information as part of PREs. Vehicle status quo, the current status of owned vehicles, such as the number of vehicles owned and their fuel-types, classes, and purchase price, is scarcely studied considering its importance. The vehicle status quo is the customer's answer to the previous vehicle purchasing process and includes every unrevealed corner of the process. Therefore, it is inevitably core information to be included, however challenging because of the scarcity of adequate data and correlation with other personal features. In this study, the vehicle status quo data is included in our analysis, which was possible because of the application of the random forest classifier (RFC). The non-rigidity of the RFC enables the highly correlated features to be analyzed for their importance. As the result, this study showed that the PREs' ranking of importance differs depending on the vehicle fuel-type. Furthermore, we analyzed the non-linear relationship between the PRE and preference towards each vehicle fuel-type with partial dependence plots.-
dc.description.abstract자동차 연료유형에 따라 그 선호도에 영향을 미치는 요인은 다르고, 이에 대한 이해를 통해 효과적인 정책 대상 도출이 가능하다. 본 연구는 친환경차 유도정책의 대상을 효율적으로 도출하기 위해 4개의 주요 자동차 연료유형(가솔린, 디젤, 전기, 수소) 각각의 선호도와 관련된 요소 차이들이 연료유형에 따라 가지는 차이를 파악하는 새로운 방법을 제시하는 것을 목적으로 한다. 선호도에 미치는 요인들의 목록은 현재 보유 중인 차량 정보를 포함하는 차량 소유자의 속성과 차량 자체의 속성을 모두 포함했다. 연료유형 별 선호도에 미치는 요인들의 중요도 비교는 컨조인트 설문조사 결과 얻은 데이터를 학습시킨 random forest classifier 모델에 게임이론을 기반으로 한 SHAP의 적용을 통해 가능했다. 각 연료유형의 요인별 섀플리 값과 이에 대한 partial dependence plot을 비교한 결과, 효과적인 정책 대상을 도출할 수 있었다. 또한, 전기차 충전 및 수소차의 수소 충전 기반시설에 대한 투자 정책도 보다 효과적으로 도출할 수 있었다. 마지막으로, SHAP 등 XAI 도구를 이용함으로써 상호연관성이 높아 기존의 방식으로 분석하기 어려웠거나, 데이터 종류가 많아 기존 모델로 정확한 분석 결과를 얻기 어려웠던 현재 보유 중인 차량 정보 등의 요인들을 동시에 고려할 수 있었다.-
dc.description.tableofcontentsChapter 1. Introduction 1
1.1 Research Background 1
1.2 Research Objectives 2
1.3 Research Outline 3
Chapter 2. Literature Review 5
2.1 Vehicle Fuel-type Preference Elements 5
2.1.1 Vehicle Related Features 7
2.1.2 Customer Related Features 8
2.2 Feature Importance Analysis for Choice Models 9
2.2.1 Conjoint Survey 10
2.2.2 Multinomial Logit Regression 11
2.2.3 Random Forest Classifier 12
2.2.4 Shapley Value 13
2.2.5 Partial Dependence Plot 13
Chapter 3. Data and Methods 15
3.1 Data Description 15
3.2 Model Description 20
Chapter 4. Results 22
4.1 Comparison with Multinomial Logit Regression 22
4.2 Element importance depending on each vehicle fuel-type 25
4.3 Effect of Major Elements 32
4.3.1 Electric Recharging and Hydrogen Refueling Infrastructure 32
4.3.2 Status Quo Vehicle Average Price and Household Income 38
4.3.3 Status Quo Vehicle Fuel-type 44
4.3.4 Age and Gender 48
4.3.5 Mileage and Fuel Cost 53
4.3.6 Policy Understanding and Technology Understanding 58
Chapter 5. Discussion 64
5.1 Key Findings and Contribution of the Research 65
5.2 Limitations and Future Research Topics 68
Bibliography 70
Appendix 1: Mixed Logit Model 76
Abstract (Korean) 78
-
dc.format.extentxi, 78-
dc.language.isoeng-
dc.publisher서울대학교 대학원-
dc.subjectVehicle Fuel-type Preference-
dc.subjectZero-emission Vehicle-
dc.subjectTarget Segmentation-
dc.subjectConjoint Survey-
dc.subjectMachine Learning-
dc.subjectShapley Additive Explanation-
dc.subject.ddc658.514-
dc.titleVehicle Fuel-type Preference Related Elements, Comparing their Importance Depending on Each Fuel-type-
dc.title.alternative자동차 연료유형 선호도와 관련된 요인들의 연료유형별 중요도 차이 비교-
dc.typeThesis-
dc.typeDissertation-
dc.contributor.AlternativeAuthorMina Kim-
dc.contributor.department공과대학 협동과정 기술경영·경제·정책전공-
dc.description.degree석사-
dc.date.awarded2023-02-
dc.identifier.uciI804:11032-000000175496-
dc.identifier.holdings000000000049▲000000000056▲000000175496▲-
Appears in Collections:
Files in This Item:

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share