Sabreena Anowar
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Sabreena Anowar has a total of 36 co-authors. Here are the top co-authors by number of shared publications:
- Naveen Eluru
University of Central Florida
16 shared publications Luis F. Miranda-Moreno
McGill University
3 shared publicationsMarianne Hatzopoulou
Department of Civil Engineering , University of Toronto , Toronto , Ontario M5S 1A4 , Canada
3 shared publicationsNowreen Keya
Univ. of Central Florida
3 shared publications- Samuel Chng
Lee Kuan Yew Centre for Innovative Cities, Singapore University of Technology and Design, Singapore 487372, Singapore
3 shared publications - Shamsunnahar Yasmin
Queensland University of Technology, School of Civil and Environmental Engineering, Brisbane, Australia
3 shared publications
Latest Publications
- Open Access
Modeling Freight Mode Choice Using Machine Learning Classifiers: A Comparative Study Using the Commodity Flow Survey (CFS) Data
This study explores the usefulness of machine learning classifiers for modeling freight mode choice. We investigate eight commonly used machine learning classifiers, namely Naive Bayes, Support Vector Machine, Artificial Neural Network, K-Nearest Neighbors, Classification and Regression Tree, Random Forest, Boosting and Bagging, along with the classical Multinomial Logit model. US 2012 Commodity Flow Survey data are used as the primary data source; we augment it with spatial attributes from secondary data sources. The performance of the classifiers is compared based on prediction accuracy results. The current research also examines the role of sample size and training-testing data split ratios on the predictive ability of the various approaches. In addition, the importance of variables is estimated to determine how the variables influence freight mode choice. The results show that the tree-based ensemble classifiers perform the best. Specifically, Random Forest produces the most accurate predictions, closely followed by Boosting and Bagging. With regard to variable importance, shipment characteristics, such as shipment distance, industry classification of the shipper and shipment size, are the most significant factors for freight mode choice decisions.
Open AccessModeling Freight Mode Choice Using Machine Learning Classifiers: A Comparative Study Using the Commodity Flow Survey (CFS) Data
This study explores the usefulness of machine learning classifiers for modeling freight mode choice. We investigate eight commonly used machine learning classifiers, namely Naive Bayes, Support Vector Machine, Artificial Neural Network, K-Nearest Neighbors, Classification and Regression Tree, Random Forest, Boosting and Bagging, along with the classical Multinomial Logit model. US 2012 Commodity Flow Survey data are used as the primary data source; we augment it with spatial attributes from secondary data sources. The performance of the classifiers is compared based on prediction accuracy results. The current research also examines the role of sample size and training-testing data split ratios on the predictive ability of the various approaches. In addition, the importance of variables is estimated to determine how the variables influence freight mode choice. The results show that the tree-based ensemble classifiers perform the best. Specifically, Random Forest produces the most accurate predictions, closely followed by Boosting and Bagging. With regard to variable importance, shipment characteristics, such as shipment distance, industry classification of the shipper and shipment size, are the most significant factors for freight mode choice decisions.
- Top Cited Papers
- 20 March 2023
- Published by Taylor & Francis in Transport Reviews
- Vol. 43 (5) , 970-996
Review of studies on public acceptability and acceptance of shared autonomous mobility services: past, present and future
The emergence of vehicle automation and its subsequent growth has led to new transport service offerings, generally known as Autonomous Mobility Services (AMS), that have the potential to replace human-operated vehicles. However, the functionalities of AMS are increasingly blurring the fine lines that currently distinguish different transport modes. For example, an autonomous shuttle bus, a form of autonomous transit, may serve a similar function as an autonomous taxi/robo-taxi, both coinciding with the concept of Shared Autonomous Mobility Services (SAMS). Even if the functionalities or operational principles are different, people may perceive sharing rides in any of these services as alike. Similarly, the absence of a human driver makes the concepts of autonomous carsharing and ridehailing similar. Hence, there is a need to review studies related to SAMS. However, few studies have attempted to perform a comprehensive review of public acceptance of SAMS. This study aims to fill this gap by reviewing studies related to public acceptability and acceptance, perception, intention to use, attitudes, mode choice and willingness constructs regarding SAMS. This review clearly distinguishes different types of SAMS while examining public’s acceptability and acceptance of SAMS across five dimensions: perception about the services, intention to choose and use those over other modal alternatives, frequency of usage and willingness constructs. Overall, the results from our review indicate the presence of heterogeneity across sub-groups regarding the adoption of SAMS. Discussing the factors affecting SAMS acceptance in a detailed manner, our study serves to provide a stocktake of the progress in this genre of research.
- 20 March 2023
- Published by Taylor & Francis in Transport Reviews
- Vol. 43 (5) , 970-996
Top Cited PapersReview of studies on public acceptability and acceptance of shared autonomous mobility services: past, present and future
The emergence of vehicle automation and its subsequent growth has led to new transport service offerings, generally known as Autonomous Mobility Services (AMS), that have the potential to replace human-operated vehicles. However, the functionalities of AMS are increasingly blurring the fine lines that currently distinguish different transport modes. For example, an autonomous shuttle bus, a form of autonomous transit, may serve a similar function as an autonomous taxi/robo-taxi, both coinciding with the concept of Shared Autonomous Mobility Services (SAMS). Even if the functionalities or operational principles are different, people may perceive sharing rides in any of these services as alike. Similarly, the absence of a human driver makes the concepts of autonomous carsharing and ridehailing similar. Hence, there is a need to review studies related to SAMS. However, few studies have attempted to perform a comprehensive review of public acceptance of SAMS. This study aims to fill this gap by reviewing studies related to public acceptability and acceptance, perception, intention to use, attitudes, mode choice and willingness constructs regarding SAMS. This review clearly distinguishes different types of SAMS while examining public’s acceptability and acceptance of SAMS across five dimensions: perception about the services, intention to choose and use those over other modal alternatives, frequency of usage and willingness constructs. Overall, the results from our review indicate the presence of heterogeneity across sub-groups regarding the adoption of SAMS. Discussing the factors affecting SAMS acceptance in a detailed manner, our study serves to provide a stocktake of the progress in this genre of research.
- Open Access
- 21 October 2022
- Published by MDPI AG in Sustainability
- Vol. 14 (20) , 13656
Understanding Shared Autonomous Vehicle Preferences: A Comparison between Shuttles, Buses, Ridesharing and Taxis
Shared autonomous vehicles (AVs) will soon be introduced in public transportation as cities and their transportation systems become ‘smarter’. This brings long-term environmental, economic and societal benefits to cities. However, shared AVs will not only need to overcome technological challenges but also prevail against social barriers for successful marketplace penetration. Hence, we proposed and investigated the acceptance of four shared AV service designs for public use in this study, namely, autonomous buses, shuttles, AV rideshares and autonomous or robo-taxis. An online survey conducted in Singapore with 734 adults found the greatest receptiveness toward the introduction of autonomous shuttles, in part due to perceptions that they will perform well and be easy to adopt. This aligns with ongoing shared AV trials where AV shuttles are mostly used. Larger autonomous buses had the second-highest acceptance. AV rideshares and taxis seem to largely appeal to the existing regular users of the conventional counterparts of these services. These results suggest that to encourage a mode switch from public transport to ridesharing and taxis, or vice versa, shared AVs need to appeal to users beyond being an automated version of existing modes. That is, shared AVs need to address an underserved or unmet transportation need or population.
- 21 October 2022
- Published by MDPI AG in Sustainability
- Vol. 14 (20) , 13656
Open AccessUnderstanding Shared Autonomous Vehicle Preferences: A Comparison between Shuttles, Buses, Ridesharing and Taxis
Shared autonomous vehicles (AVs) will soon be introduced in public transportation as cities and their transportation systems become ‘smarter’. This brings long-term environmental, economic and societal benefits to cities. However, shared AVs will not only need to overcome technological challenges but also prevail against social barriers for successful marketplace penetration. Hence, we proposed and investigated the acceptance of four shared AV service designs for public use in this study, namely, autonomous buses, shuttles, AV rideshares and autonomous or robo-taxis. An online survey conducted in Singapore with 734 adults found the greatest receptiveness toward the introduction of autonomous shuttles, in part due to perceptions that they will perform well and be easy to adopt. This aligns with ongoing shared AV trials where AV shuttles are mostly used. Larger autonomous buses had the second-highest acceptance. AV rideshares and taxis seem to largely appeal to the existing regular users of the conventional counterparts of these services. These results suggest that to encourage a mode switch from public transport to ridesharing and taxis, or vice versa, shared AVs need to appeal to users beyond being an automated version of existing modes. That is, shared AVs need to address an underserved or unmet transportation need or population.
- 1 December 2021
- Published by Elsevier in Case Studies on Transport Policy
- Vol. 9 (4) , 1542-1552
To embrace or not to embrace? Understanding public’s dilemma about autonomous mobility services: A case study of Singapore
- 1 December 2021
- Published by Elsevier in Case Studies on Transport Policy
- Vol. 9 (4) , 1542-1552
To embrace or not to embrace? Understanding public’s dilemma about autonomous mobility services: A case study of Singapore
- Open Access
- 24 May 2021
- Published by Taylor & Francis in Transportation Planning and Technology
- Vol. 44 (5) , 543-559
Modeling freight mode choice using machine learning classifiers: a comparative study using Commodity Flow Survey (CFS) data
This study explores the usefulness of machine learning classifiers for modeling freight mode choice. We investigate eight commonly used machine learning classifiers, namely Naïve Bayes, Support Vector Machine, Artificial Neural Network, K-Nearest Neighbors, Classification and Regression Tree, Random Forest, Boosting and Bagging, along with the classical Multinomial Logit model. US 2012 Commodity Flow Survey data are used as the primary data source; we augment it with spatial attributes from secondary data sources. The performance of the classifiers is compared based on prediction accuracy results. The current research also examines the role of sample size and training-testing data split ratios on the predictive ability of the various approaches. In addition, the importance of variables is estimated to determine how the variables influence freight mode choice. The results show that the tree-based ensemble classifiers perform the best. Specifically, Random Forest produces the most accurate predictions, closely followed by Boosting and Bagging. With regard to variable importance, shipment characteristics, such as shipment distance, industry classification of the shipper and shipment size, are the most significant factors for freight mode choice decisions.
- 24 May 2021
- Published by Taylor & Francis in Transportation Planning and Technology
- Vol. 44 (5) , 543-559
Open AccessModeling freight mode choice using machine learning classifiers: a comparative study using Commodity Flow Survey (CFS) data
This study explores the usefulness of machine learning classifiers for modeling freight mode choice. We investigate eight commonly used machine learning classifiers, namely Naïve Bayes, Support Vector Machine, Artificial Neural Network, K-Nearest Neighbors, Classification and Regression Tree, Random Forest, Boosting and Bagging, along with the classical Multinomial Logit model. US 2012 Commodity Flow Survey data are used as the primary data source; we augment it with spatial attributes from secondary data sources. The performance of the classifiers is compared based on prediction accuracy results. The current research also examines the role of sample size and training-testing data split ratios on the predictive ability of the various approaches. In addition, the importance of variables is estimated to determine how the variables influence freight mode choice. The results show that the tree-based ensemble classifiers perform the best. Specifically, Random Forest produces the most accurate predictions, closely followed by Boosting and Bagging. With regard to variable importance, shipment characteristics, such as shipment distance, industry classification of the shipper and shipment size, are the most significant factors for freight mode choice decisions.
- 1 February 2020
- Published by American Society of Civil Engineers (ASCE) in Journal of Transportation Engineering, Part A: Systems
- Vol. 146 (2)
Fusing Freight Analysis Framework and Transearch Data: Econometric Data Fusion Approach with Application to Florida
A major hurdle in freight demand modeling has always been the lack of adequate data on freight movements for different industry sectors for planning applications. Both Freight Analysis Framework (FAF) and Transearch (TS) databases contain annualized commodity flow data. However, the representations of commodity flows in the two databases are inherently different. FAF flows represent estimated transportation network flows, whereas TS flows represent production–consumption commodity flows. This study aims to develop a fused database from FAF and TS to determine transportation network flows at a fine spatial resolution (county level) while accommodating production and consumption behavioral trends (provided by TS). To this end, a joint econometric model framework embedded within a network flow approach and grounded in a maximum-likelihood technique is formulated to estimate county-level commodity flows. The algorithm is implemented for the commodity flow information from 2012 FAF data and 2011 TS databases to generate transportation network flows for 67 counties in Florida. The proposed approach may be able to circumvent the need to purchase expensive TS databases in the future.
- 1 February 2020
- Published by American Society of Civil Engineers (ASCE) in Journal of Transportation Engineering, Part A: Systems
- Vol. 146 (2)
Fusing Freight Analysis Framework and Transearch Data: Econometric Data Fusion Approach with Application to Florida
A major hurdle in freight demand modeling has always been the lack of adequate data on freight movements for different industry sectors for planning applications. Both Freight Analysis Framework (FAF) and Transearch (TS) databases contain annualized commodity flow data. However, the representations of commodity flows in the two databases are inherently different. FAF flows represent estimated transportation network flows, whereas TS flows represent production–consumption commodity flows. This study aims to develop a fused database from FAF and TS to determine transportation network flows at a fine spatial resolution (county level) while accommodating production and consumption behavioral trends (provided by TS). To this end, a joint econometric model framework embedded within a network flow approach and grounded in a maximum-likelihood technique is formulated to estimate county-level commodity flows. The algorithm is implemented for the commodity flow information from 2012 FAF data and 2011 TS databases to generate transportation network flows for 67 counties in Florida. The proposed approach may be able to circumvent the need to purchase expensive TS databases in the future.
- 14 March 2019
- Published by Taylor & Francis in International Journal of Sustainable Transportation
- Vol. 14 (7) , 544-553
Disaggregate level simulation of bus transit emissions in a large urban region
In this study, we demonstrate the development of a methodology for simulating transit bus ridership and greenhouse gas emissions (in CO2 equivalent) across a network of 200 bus lines in the city of Montreal, Canada. The current study estimates ridership and emissions at a micro level, that is, at the disaggregate stop level for 24 h of a typical weekday for all the bus lines (and their trips) in the bus network of Montreal. The disaggregate level simulation process incorporates for specificities such as vehicle type, age, fuel, and passenger load. The micro-simulation platform developed has three embedded modules: (1) stop level boarding and alighting, (2) bus occupancy determination, and (3) disaggregate emission estimation. Our simulation allows us to estimate emissions for individual buses while running as well as while idling at bus stops. The proposed model system provides measures of per capita emissions (total emissions/total boardings), per km emissions (total emissions/total kilometers traveled), and per person km emissions (total emissions/total person kilometers traveled on bus transit). To further illustrate the applicability of the proposed study, we conducted several policy scenario analyses. We investigated the effect of a 20% system-wide increase in ridership and observed a 1.1% increase in total emissions and a 13% decrease in per capita emissions. We estimated the effects of decreasing the frequency of low occupancy buses and increasing the frequency of high occupancy buses as well and observed that these frequency changes are associated with proportional changes in emissions.
- 14 March 2019
- Published by Taylor & Francis in International Journal of Sustainable Transportation
- Vol. 14 (7) , 544-553
Disaggregate level simulation of bus transit emissions in a large urban region
In this study, we demonstrate the development of a methodology for simulating transit bus ridership and greenhouse gas emissions (in CO2 equivalent) across a network of 200 bus lines in the city of Montreal, Canada. The current study estimates ridership and emissions at a micro level, that is, at the disaggregate stop level for 24 h of a typical weekday for all the bus lines (and their trips) in the bus network of Montreal. The disaggregate level simulation process incorporates for specificities such as vehicle type, age, fuel, and passenger load. The micro-simulation platform developed has three embedded modules: (1) stop level boarding and alighting, (2) bus occupancy determination, and (3) disaggregate emission estimation. Our simulation allows us to estimate emissions for individual buses while running as well as while idling at bus stops. The proposed model system provides measures of per capita emissions (total emissions/total boardings), per km emissions (total emissions/total kilometers traveled), and per person km emissions (total emissions/total person kilometers traveled on bus transit). To further illustrate the applicability of the proposed study, we conducted several policy scenario analyses. We investigated the effect of a 20% system-wide increase in ridership and observed a 1.1% increase in total emissions and a 13% decrease in per capita emissions. We estimated the effects of decreasing the frequency of low occupancy buses and increasing the frequency of high occupancy buses as well and observed that these frequency changes are associated with proportional changes in emissions.
- 12 February 2019
- Published by Taylor & Francis in Transportmetrica A: Transport Science
- Vol. 15 (2) , 1214-1246
Regret minimization based joint econometric model of mode choice and departure time: a case study of university students in Toronto, Canada
With the objective of enhancing our understanding of student travel patterns, we examine their mode and departure time choice for discretionary trip purposes. In our study, we hypothesize that students are likely to consider these joint choices or interconnected decisions in a sequence (mode first, departure time second and departure time first, mode second), even if the time difference between these decisions is infinitesimally small. A latent segmentation approach is proposed that allows for the incorporation of choice sequences simultaneously. Within this framework, mode and departure time decisions are examined using random utility and random regret decision framework. This is the first application of such a joint sequential system using a regret-based decision rule. Our results indicate that students are more likely to choose their departure time first and then mode. A policy exercise is also conducted to demonstrate the applicability of the proposed model system.
- 12 February 2019
- Published by Taylor & Francis in Transportmetrica A: Transport Science
- Vol. 15 (2) , 1214-1246
Regret minimization based joint econometric model of mode choice and departure time: a case study of university students in Toronto, Canada
With the objective of enhancing our understanding of student travel patterns, we examine their mode and departure time choice for discretionary trip purposes. In our study, we hypothesize that students are likely to consider these joint choices or interconnected decisions in a sequence (mode first, departure time second and departure time first, mode second), even if the time difference between these decisions is infinitesimally small. A latent segmentation approach is proposed that allows for the incorporation of choice sequences simultaneously. Within this framework, mode and departure time decisions are examined using random utility and random regret decision framework. This is the first application of such a joint sequential system using a regret-based decision rule. Our results indicate that students are more likely to choose their departure time first and then mode. A policy exercise is also conducted to demonstrate the applicability of the proposed model system.
- 8 February 2019
- Published by American Chemical Society (ACS) in Environmental Science & Technology
- Vol. 53 (5) , 2799-2810
Within-City Spatial Variations in Multiple Measures of PM2.5 Oxidative Potential in Toronto, Canada
Few studies have characterized within-city spatial variations in the oxidative potential of fine particulate air pollution (PM2.5). In this study, we evaluated multiple measures of PM2.5 oxidative potential across Toronto, Canada (2016-2017) including glutathione/ascorbate-related oxidative potential (OPGSH and OPAA) and dithiothreitol depletion (OPDTT). Integrated 2-week samples were collected from 67 sites in summer and 42 sites in winter. Multivariable linear models were developed to predict OP based on various land use/traffic factors and PM2.5 metals and black carbon were also examined. All three measures of PM2.5 oxidative potential varied substantially across Toronto. OPAA and OPDTT were primarily associated with traffic-related components of PM2.5 (i.e. Fe, Cu, and black carbon) whereas OPGSH was not a strong marker for traffic during either season. During summer, multivariable models performed best for OPAA (R2CV=0.48) followed by OPDTT (R2CV=0.32) and OPGSH (R2CV=0.22). During winter, model performance was best for OPDTT (R2CV=0.55) followed by OPGSH (R2CV=0.50) and OPAA (R2CV=0.23). Model parameters varied between seasons and between-season differences in PM2.5 mass concentrations were weakly/moderately correlated with seasonal differences in OP. Our findings highlight substantial within-city variations in PM2.5 oxidative potential. More detailed information is needed on local sources of air pollution to improve model performance.
- 8 February 2019
- Published by American Chemical Society (ACS) in Environmental Science & Technology
- Vol. 53 (5) , 2799-2810
Within-City Spatial Variations in Multiple Measures of PM2.5 Oxidative Potential in Toronto, Canada
Few studies have characterized within-city spatial variations in the oxidative potential of fine particulate air pollution (PM2.5). In this study, we evaluated multiple measures of PM2.5 oxidative potential across Toronto, Canada (2016-2017) including glutathione/ascorbate-related oxidative potential (OPGSH and OPAA) and dithiothreitol depletion (OPDTT). Integrated 2-week samples were collected from 67 sites in summer and 42 sites in winter. Multivariable linear models were developed to predict OP based on various land use/traffic factors and PM2.5 metals and black carbon were also examined. All three measures of PM2.5 oxidative potential varied substantially across Toronto. OPAA and OPDTT were primarily associated with traffic-related components of PM2.5 (i.e. Fe, Cu, and black carbon) whereas OPGSH was not a strong marker for traffic during either season. During summer, multivariable models performed best for OPAA (R2CV=0.48) followed by OPDTT (R2CV=0.32) and OPGSH (R2CV=0.22). During winter, model performance was best for OPDTT (R2CV=0.55) followed by OPGSH (R2CV=0.50) and OPAA (R2CV=0.23). Model parameters varied between seasons and between-season differences in PM2.5 mass concentrations were weakly/moderately correlated with seasonal differences in OP. Our findings highlight substantial within-city variations in PM2.5 oxidative potential. More detailed information is needed on local sources of air pollution to improve model performance.