서지주요정보
Unveiling functional regions through travel dynamics = 통행 행태 기반 기능 지역 연구
서명 / 저자 Unveiling functional regions through travel dynamics = 통행 행태 기반 기능 지역 연구 / Sujin Lee.
발행사항 [대전 : 한국과학기술원, 2024].
Online Access 원문보기 원문인쇄

소장정보

등록번호

8043051

소장위치/청구기호

학술문화관(도서관)2층 학위논문

DGT 24006

휴대폰 전송

도서상태

이용가능(대출불가)

사유안내

반납예정일

리뷰정보

초록정보

Understanding functional regions resulting from interactions between people and regions is crucial for effective city planning and transportation management. Addressing potential deviations from planned functional regions, this study focuses on identifying how people actually utilize regions based on travel demand patterns. We propose a research framework consisting of three phases—extraction, validation, and application—using multi-modal travel demand datasets covering buses, taxis, and probe vehicles over a long-term period. In the extraction phase, regional clusters are identified based on similar temporal travel demand patterns across multiple transportation modes within a year using t-distributed stochastic neighbor embedding and K-Means clustering methods. The validation phase aims to prove the distinctiveness of regional clusters on both collective and individual scales. Spatial characteristics, considering the built environment and socio-demography, interpret each regional cluster using extreme gradient boosting and Shapley additive explanations, labeling regional clusters with representative functions. On the individual scale, differences in travel behavioral indices across functional regions, in this case, the revisit interval and stay duration, are analyzed. In the application phase, functional regions, as indicators incorporating spatial-temporal features for each region, contribute to enhancing future travel demand forecasting. As a case study, our research framework is applied to the urban area of Daejeon Metropolitan City, South Korea. The study identifies six distinct functional regions, specifically residential areas (proximity to the city center and outskirts), industrial zones, business/education/ research districts, commercial centers, and mixed-functional regions. It also confirms that individual travel behaviors vary across functional regions and that embedding functional region variables can enhance the predictability of demand for city services. This study provides a detailed understanding of the spatio-temporal dynamics linked to functional regions and offers valuable insights into functional regions illuminated by its analysis of travel demand patterns. These insights can support informed decisions in urban planning and transportation management.

도시의 효율적 운영을 위해서는 지역과 사람 간 상호 작용으로 사람들이 실제로 각 지역을 어떤 목적으로 활용하는지, 기능 지역에 대한 이해가 중요하다. 본 연구는 버스, 택시 및 프로브 차량을 포함한 다양한 교통수단의 통행 기록을 활용하여 기능 지역에 대한 3단계 연구 프레임 워크를 제안한다. 1단계-기능 지역 추출 단계에서는 다중 교통수단의 통행 기록에서 얻은 시간적 수요 패턴의 유사성을 기반으로 t-분포 확률적 이웃 임베딩과 K-Means 클러스터링 방법을 사용하여 지역 클러스터를 식별한다. 2단계-검증 단계는 지역 클러스터의 특징점을 집합적 및 개별적 규모에서 입증하는 것을 목표로 한다. 집합적 규모에서의 검증은 건축 환경 및 사회인구학을 고려한 공간 특성을 활용하여 Extreme Gradient Boosting과 Shapley Additive Explanations을 사용하여 각 지역 클러스터를 해석하고 지역 클러스터에 대표적 기능을 부여하여 기능 지역을 도출한다. 개별적 규모에서는 재방문 간격 및 체류 기간과 같은 개별 통행 행태 지표를 활용하여 기능 지역 간 지표의 차이를 분석한다. 기능 지역은 각 지역의 시공간적 특징을 내포한 변수로, 3단계-응용 단계에서 단기 통행 수요 예측 성능을 향상시키는 데에 활용한다. 본 연구 프레임 워크를 검증하기 위해 대전광역시를 대상으로 시범 분석을 진행하였으며, 주거 지역, 공업 지역, 업무/교육/연구 지역, 상업 지역 및 혼합 기능적 지역을 포함한 6개의 기능 지역을 도출하였다. 또한, 개별 통행 행태가 기능 지역별로 다르게 분포하며, 기능 지역 변수의 사용이 통행 수요 예측 성능 개선으로 이어지는 것을 확인하였다. 이에 따라 본 연구에서 제안한 연구 프레임 워크는 기능 지역의 시공간적 영향에 대한 이해를 제공하며, 이러한 통찰력은 도시 계획 및 교통 관리에 체계적인 의사 결정을 지원할 수 있을 것으로 사료된다.

서지기타정보

서지기타정보
청구기호 {DGT 24006
형태사항 v,138 p. : 삽도 ; 30 cm
언어 영어
일반주기 저자명의 한글표기 : 이수진
지도교수의 영문표기 : Kitae Jang
지도교수의 한글표기 : 장기태
Including appendix
학위논문 학위논문(박사) - 한국과학기술원 : 조천식모빌리티대학원,
서지주기 References : p. 119-130
주제 Functional region
Travel behavior
Travel demand
Land use
Built environment
Socio-demography
Clustering analysis
Spatial analysis
Travel demand prediction
기능 지역
통행 행태
통행 수요
토지 이용
건축 환경
사회 인구학
클러스터링 분석
공간 분석
통행 수요 예측
QR CODE

책소개

전체보기

목차

전체보기

이 주제의 인기대출도서

Interactions between regions and people via activity

Dissertation organization

Representative keywords used in studies related to functional regions

Distribution of studies on functional regions across data sources

Remote sensingimages - satellite images (Daejeon, Kakao map)

General research flow in studies identifying functional regions

Conceptual research framework

Location of Daejeon Metropolitan City in South Korea

Modal share rates by urban transportation mode (for the year 2018)

Data structure of processed bus trip records (smart card system)

Temporal patterns ofbus demand over month, day of week, and hour

Data structure of taxi trip records (taxi fare meter system)

Temporal patterns of taxi demand over month, day of week, and hour

Dedicated short-range communication (DSRC) system in Daejeon

Data structure ofDSRC-based vehicle trajectory data in Daejeon (abstract)

Temporal patterns ofprobe vehicle demand over month,dayofweek, and hour

Spatial datasets in Daejeon Metropolitan City, South Korea

Functional region extraction and collective-scale inference

Concept of stochastic neighbor embedding

Dimension reduction on the temporal axis of the travel demand matrices for buses and taxis

Concepts of (a) K-Means clustering and (b) agglomerative hierarchical clustering

Clustering performances of candidate clustering methods

List of spatial datasets and variables

Heatmap of the correlations among spatial variables

Concept of an explainable model to interpret the black-box model behavior

Number of cells across regional clusters

Demand patterns of buses and taxis across regional clusters over months

Demand patterns of buses and taxis across regional clusters over days of the week

Demand patterns of buses and taxis across regional clusters over days of the week and hours oftheday

Classification performance of XGBoost

Geographical distribution and relationships ofspatial attributes across regional clusters

Geographical distribution and relationships ofspatial attributes for regional cluster 1

Geographical distribution and relationships ofspatial attributes for regional cluster 2

Geographical distribution and relationships ofspatial attributes for regional cluster 3

Geographical distribution and relationships ofspatial attributes for regional cluster

Geographical distribution and relationships ofspatial attributes for regional cluster 5

Geographical distribution and relationships ofspatial attributes for regional cluster 6

Relationships among regional clusters derived from three different modal

Temporal demand patterns across regional clusters according to a single transportation mode

Geographical distribution ofregional clusters according to a singletransportation mode

Estimated multinomial logit regression coefficients

Scatter plot of SHAP values and proportions of residential buildings

Travel demand scale across regional clusters

Concept of measuring the revisit interval and stay duration at the individua level

Buffer areas for the next departure location

Distance between two cumulative distributions according to the Smirnov test

Revisit interval distribution across functional regions

Visualization ofSmirnov statistics of revisit intervals

Probability of revisit intervals at 24 hours

Distribution ofstay durations across functional regions

Visualization of Smirnov statistics ofstay durations

Prediction horizon in time-series data

Travel demand prediction using a Temporal Fusion Transformer network

Model variants using different combinations ofinput features

Prediction performances across model variants

Prediction performances of model variants across functional regions

Sample prediction outcomes by model variants and functional regions

Sample prediction outcomes by model variants and functional regions (contd.)

Sample prediction outcomes by model variants and functional regions (contd.)

Feature importancebyinputtypes across model variants

Prediction performance ofTFT-Cluster depending on time features

Research summary

Principal component analysis of temporal travel demand - month

Principal componentanalysis oftemporal travel demand -day of week

Smirnov Statistics ofRevisit Intervals across Functional Regions

Smirnov Statistics of Stay Durations across Functional Regions