DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Jang, Kitae | - |
dc.contributor.advisor | 장기태 | - |
dc.contributor.author | Gu, Eunmo | - |
dc.date.accessioned | 2018-06-20T06:25:11Z | - |
dc.date.available | 2018-06-20T06:25:11Z | - |
dc.date.issued | 2017 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=675501&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/243507 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 조천식녹색교통대학원, 2017.2,[iii, 41 p. :] | - |
dc.description.abstract | Buses are one of the most useful public transportation in the urban area. The most important factor in assessing the service quality of buses is the predictability of travel time. When planning one's itinerary, the main criteria for choosing transportation mode is anticipated travel time to be taken. Previous studies regarding bus travel time prediction did not take into account characteristics depending on the predicted time. In addition, recent development of ICT has collected bus traffic records from the bus information system, and it is easy to predict the travel time based on the data. As a result of analyzing the attributes of bus travel time through bus traffic record, it is shown that route based prediction is more proper than link-node for bus travel time. Based on this, the method of predicting each short-term and long-term travel time is proposed by applying artificial neural network. Long-term forecasting is based on self-organizing clustering, one of the unsupervised learning algorithm by using historical data. Short-term forecasting is based on neural network analysis for supervised learning algorithm by using both real and historical data. In order to verify the proposed framework, An empirical analysis is conducted by selecting one direct route and one trunk route from the suburb area to the central business district through a median bus priority lane. As a result, the average absolute error percentage was less than 10%, and the final results of predicting travel time with accuracy of 90% or more were derived. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | Bus Travel Time | - |
dc.subject | Time Prediction | - |
dc.subject | Machine Learning | - |
dc.subject | Clustering | - |
dc.subject | Artificial Neural Network | - |
dc.subject | 버스통행시간 | - |
dc.subject | 시간예측 | - |
dc.subject | 기계학습 | - |
dc.subject | 클러스터링 | - |
dc.subject | 인공신경망 | - |
dc.title | Attributes and prediction for bus travel time | - |
dc.title.alternative | 버스 통행시간의 특성 및 예측 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :조천식녹색교통대학원, | - |
dc.contributor.alternativeauthor | 구은모 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.