Melrose M Pan | University of Arizona (original) (raw)

Papers by Melrose M Pan

XXI PANAM Lima 2021: XXI Pan American Conference of Transport and Logistics, 2021

In recent years, many transportation agencies such as the Pima County Department of Transportatio... more In recent years, many transportation agencies such as the Pima County Department of Transportation in Tucson, Arizona, have subscribed to traffic data services from StreetLight Data Inc. (hereafter referred to as StreetLight). Streetlight collects data from various non-traditional survey data-based sources, mostly from location-based services (LBS). As the usage of LBS data becomes more prevalent in transportation planning, the question of using it for more complicated tasks, such as providing OD matrices for urban areas is being raised. The most outstanding question is how reliable the StreetLight OD demand is for traffic simulation. To provide more in-depth insight into the quality and reliability of the StreetLight data, we propose a framework combining qualitative and quantitative techniques to validate the travel demand. A dynamic traffic assignment (DTA) approach was utilized to validate whether or not the demand reflects actual trip patterns by comparing the simulation and actual traffic flow. The results show that StreetLight OD data can represent the general mobility patterns since the temporal and spatial patterns are reasonable. However, StreetLight OD data overestimated the actual traffic. The mean absolute percentage difference is 40.8% compared with the screen line counts collected from permanent stations. These considerations should be a part of any future use of StreetLight data for traffic simulation.

XXI PANAM Lima 2021: XXI Pan American Conference of Transport and Logistics, 2021

Technology-driven traffic demand management has emerged as a promising concept for congestion mit... more Technology-driven traffic demand management has emerged as a promising concept for congestion mitigation in recent years, among which school carpooling could receive the particular benefit if applied to an entire region. To aid in the decision-making process for the funding of such a program, we provide a method to quantify the benefits using a simulation-based dynamic traffic assignment (DTA.) modeling approach. The research questions for this paper are a. How can we increase parents' engagement levels in a school carpooling program? b. How much travel time can we theoretically save using this type of demand strategy? To help answer these questions, surveys were designed and distributed through Amazon Mechanical Turk to understand the relationship between participation rates, occupancy, and factors such as parents' demographics, kids' age, and schools' marketing strategy. Using the information gathered on these relationships, the new demand for school carpooling implementation can be estimated. A case study in the Marana school district containing 26 schools in Pima County, Arizona, was studied. We conducted three types of analyses to evaluate the travel time savings by running the DTA simulation with the original and new demand. First, link-level travel time saving was analyzed. Second, we studied the sensitivity of regional travel time saved concerning a range of participation and occupancy rates to explore the potential of school carpooling programs. Third, we assessed the annual county-wide monetary savings if we expanded the school carpooling program to the whole Pima County to evaluate the overall performance of scenarios. We found that the distribution of travel time improvement among individual links in the Marana region would have a more consistent and significant improvement with increased participation rate and occupancy. We also found that promoting the program towards and targeting families that live farther away from schools would attract more participation.

第十二届中国智能交通年会大会论文集, 2017

网络舆情是对交通数据来源的一个有力补充,它常常包含着事件的内容,致因分析以及发布者的个人情感倾向等更丰富的内容.如今移动互联网群体的数量日趋庞大,自媒体顺势而生,成为了新媒体的中坚力量.本文以微... more 网络舆情是对交通数据来源的一个有力补充,它常常包含着事件的内容,致因分析以及发布者的个人情感倾向等更丰富的内容.如今移动互联网群体的数量日趋庞大,自媒体顺势而生,成为了新媒体的中坚力量.本文以微博社交平台为主要数据来源,利用网络爬虫技术实时获取与城市交通系统相关的一系列文本数据;通过对中文语义库的二次开发实现对城市交通问题的事件特征提取与致因挖掘,同时研究文本背后的情感特征,最终开发出一套完整的以微博文本为研究对象的交通网络舆情数据收集——数据分析——特征提取——情感分析的方法论.

XXI PANAM Lima 2021: XXI Pan American Conference of Transport and Logistics, 2021

In recent years, many transportation agencies such as the Pima County Department of Transportatio... more In recent years, many transportation agencies such as the Pima County Department of Transportation in Tucson, Arizona, have subscribed to traffic data services from StreetLight Data Inc. (hereafter referred to as StreetLight). Streetlight collects data from various non-traditional survey data-based sources, mostly from location-based services (LBS). As the usage of LBS data becomes more prevalent in transportation planning, the question of using it for more complicated tasks, such as providing OD matrices for urban areas is being raised. The most outstanding question is how reliable the StreetLight OD demand is for traffic simulation. To provide more in-depth insight into the quality and reliability of the StreetLight data, we propose a framework combining qualitative and quantitative techniques to validate the travel demand. A dynamic traffic assignment (DTA) approach was utilized to validate whether or not the demand reflects actual trip patterns by comparing the simulation and actual traffic flow. The results show that StreetLight OD data can represent the general mobility patterns since the temporal and spatial patterns are reasonable. However, StreetLight OD data overestimated the actual traffic. The mean absolute percentage difference is 40.8% compared with the screen line counts collected from permanent stations. These considerations should be a part of any future use of StreetLight data for traffic simulation.

XXI PANAM Lima 2021: XXI Pan American Conference of Transport and Logistics, 2021

Technology-driven traffic demand management has emerged as a promising concept for congestion mit... more Technology-driven traffic demand management has emerged as a promising concept for congestion mitigation in recent years, among which school carpooling could receive the particular benefit if applied to an entire region. To aid in the decision-making process for the funding of such a program, we provide a method to quantify the benefits using a simulation-based dynamic traffic assignment (DTA.) modeling approach. The research questions for this paper are a. How can we increase parents' engagement levels in a school carpooling program? b. How much travel time can we theoretically save using this type of demand strategy? To help answer these questions, surveys were designed and distributed through Amazon Mechanical Turk to understand the relationship between participation rates, occupancy, and factors such as parents' demographics, kids' age, and schools' marketing strategy. Using the information gathered on these relationships, the new demand for school carpooling implementation can be estimated. A case study in the Marana school district containing 26 schools in Pima County, Arizona, was studied. We conducted three types of analyses to evaluate the travel time savings by running the DTA simulation with the original and new demand. First, link-level travel time saving was analyzed. Second, we studied the sensitivity of regional travel time saved concerning a range of participation and occupancy rates to explore the potential of school carpooling programs. Third, we assessed the annual county-wide monetary savings if we expanded the school carpooling program to the whole Pima County to evaluate the overall performance of scenarios. We found that the distribution of travel time improvement among individual links in the Marana region would have a more consistent and significant improvement with increased participation rate and occupancy. We also found that promoting the program towards and targeting families that live farther away from schools would attract more participation.

第十二届中国智能交通年会大会论文集, 2017

网络舆情是对交通数据来源的一个有力补充,它常常包含着事件的内容,致因分析以及发布者的个人情感倾向等更丰富的内容.如今移动互联网群体的数量日趋庞大,自媒体顺势而生,成为了新媒体的中坚力量.本文以微... more 网络舆情是对交通数据来源的一个有力补充,它常常包含着事件的内容,致因分析以及发布者的个人情感倾向等更丰富的内容.如今移动互联网群体的数量日趋庞大,自媒体顺势而生,成为了新媒体的中坚力量.本文以微博社交平台为主要数据来源,利用网络爬虫技术实时获取与城市交通系统相关的一系列文本数据;通过对中文语义库的二次开发实现对城市交通问题的事件特征提取与致因挖掘,同时研究文本背后的情感特征,最终开发出一套完整的以微博文本为研究对象的交通网络舆情数据收集——数据分析——特征提取——情感分析的方法论.