Activity-based Travel Demand Research Articles

Exploring the Spatial Transferability of FEATHERS – An Activity Based Travel Demand Model – For Ho Chi Minh City, Vietnam

This paper presents descriptive analyses to investigate the transferability of FEATHERS, an activity-based travel demand model developed for Flanders (Belgium), a region in a developed country to Ho Chi Minh City (Vietnam), a metropolitan area of a developing country using personal and household travel survey data. The transferability was tested by using indicators which represent individuals’ activity and travel behaviors. The results confirm the transferability of FEATHERS modelling structure to Ho Chi Minh City (HCMC), but all sub-models require recalibration due to the different cultural settings between the two study areas. The differences in transport mode options, land use patterns and individuals’ location choice preferences make transport mode and location choice the least transferable models.

Analysis of Activity Location and Trip Mode Choice – A Study on Hierarchical Ordering

This paper focusses on preliminary analyses conducted to investigate the hierarchical decision-making structure of location and mode choice problems for non-work/school activities. The travel behaviour of two streams of individuals are studied separately. Nested logit models are tested using full information maximum likelihood method after estimating independent models for location and mode choice. The results show that individuals are more inclined to change their destination rather than their mode. The architectural framework of TASHA, the operational agent-based activity-based travel demand model of the Greater Toronto and Hamilton Area, will be redesigned in the light of these results and further analyses.

GPU-based Parallel Computing for Activity-based Travel Demand Models

Activity-based travel demand models (ABMs) are gaining popularity in the field of traffic modeling because of their high level of detail compared to traditional travel demand models. Due to this, however, ABMs have high computational requirements, making ABMs hard to use for analysis and optimization purposes. We address this problem by relying on the concept of parallel computing using a computer’s graphics processing unit (GPU). To illustrate the potential of GPU computing for ABM, we present a pilot study in which we compare the observed computation time of an ABM GPU implementation that we built using NVIDIA’s CUDA framework with similar, non-parallel implementations. We conclude that speed-ups up to a factor 50 can be realized, enabling the use of ABMs both for fast analysis of scenarios and for optimization purposes.

Forecasting current and next trip purpose with social media data and Google Places

Trip purpose is crucial to travel behavior modeling and travel demand estimation for transportation planning and investment decisions. However, the spatial-temporal complexity of human activities makes the prediction of trip purpose a challenging problem. This research, an extension of work by Ermagun et al. (2017) and Meng et al. (2017), addresses the problem of predicting both current and next trip purposes with both Google Places and social media data. First, this paper implements a new approach to match points of interest (POIs) from the Google Places API with historical Twitter data. Therefore, the popularity of each POI can be obtained. Additionally, a Bayesian neural network (BNN) is employed to model the trip dependence on each individual’s daily trip chain and infer the trip purpose. Compared with traditional models, it is found that Google Places and Twitter information can greatly improve the overall accuracy of prediction for certain activities, including “EatOut”, “Personal”, “Recreation” and “Shopping”, but not for “Education” and “Transportation”. In addition, trip duration is found to be an important factor in inferring activity/trip purposes. Further, to address the computational challenge in the BNN, an elastic net is implemented for feature selection before the classification task. Our research can lead to three types of possible applications: activity-based travel demand modeling, survey labeling assistance, and online recommendations.

An Integrated, Validated, and Applied Activity-Based Dynamic Traffic Assignment Model for the Baltimore-Washington Region

The paper discusses the integration process and initial applications of a new model for the Baltimore-Washington region that integrates an activity-based travel demand model (ABM) with a dynamic traffic assignment (DTA) model. Specifically, the integrated model includes InSITE, an ABM developed for the Baltimore Metropolitan Council, and DTALite, a mesoscopic DTA model. The integrated model simulates the complete daily activity choices of individuals residing in the model region, including long-term choices, such as workplace location; daily activity patterns, including joint household activities and school escorting; activity location choices; time-of-day choices; mode choices; and route choices. The paper describes the model development and integration approach, including modeling challenges, such as the need to maintain consistency between the ABM and DTA models in terms of temporal and spatial resolution, and practical implementation issues, such as managing model run time and ensuring sufficient convergence of the model. The integrated model results have been validated against observed daily traffic volumes and vehicle-miles traveled (VMT) for various functional classes. A land-use change scenario that analyzes the redevelopment of the Port Covington area in Baltimore is applied and compared with the baseline scenario. The validation and application results suggest that the integrated model outperforms a static assignment-based ABM and could capture behavioral changes at much finer time resolutions.

What about the dynamics in daily travel mode choices? A dynamic discrete choice approach for tour-based mode choice modelling

The paper presents a heteroskedastic dynamic discrete choice (HDDC) model for tour-based mode choices modelling with an empirical investigation of university students' daily mode choices in Toronto. The reality of connected trips and resulting constrained mode choices are captured through the HDDC framework that is suitable for fitting in an activity-based travel demand modelling system. Data from a web-based travel survey of the students of four universities in Toronto are used. The empirical model highlights the importance of capturing the dynamics in tour-based mode choices modelling. The dynamic model reveals that students' sensitivity to cost vary by trips of the day, while their sensitivity to travel time remains stable. Results of this investigation have policy implications and the proposed methodology has applications in activity-based travel demand modelling.

Learning Daily Activity Sequences of Population Groups using Random Forest Theory

The choice of daily activity sequences differs between individuals based on their socio-demographic characteristics and their health and/or mobility status. The aim of this paper is to provide an improved methodology for learning and modeling the daily activity engagement patterns of individuals using a state-of-the-art machine learning algorithm. The dependencies between activity type, activity frequency, activity sequence, and socio-demographic characteristics of individuals are taken into account by employing a random forest model. In order to capture the heterogeneity and diversity among the predictor variables, we employed two different methods for split selection in the random forest algorithm: Classification and Regression Tree (CART) and curvature search. These two methods were examined under two different layer settings. In the first setting, the algorithm grows trees using all alternative predictor variables, whereas in the second setting the importance of the predictor variables is estimated and then the algorithm grows trees using only high-score predictor variables. The models were applied to time use data from the large Halifax Space-Time Activity Research (STAR) household travel diary survey. We evaluated the estimation accuracy of the proposed models using confusion matrix, transition matrix, and sequential alignment techniques. Results show that the random forest model with CART split selection using the first layer setting has the best accuracy in replicating activity agendas and activity sequences of individuals. The results of this paper are expected to be implemented within the activity-based travel demand model, Scheduler for Activities, Locations, and Travel (SALT).

2445 - Multifaceted Transportation Performance Integrating Mobility, Safety, Air Quality, Physical Activity and Public Health

Background While mobility, safety, and environmental friendliness have been major interests in transportation systems, recently people have started to pay more attention to public health related to transportation. Emphasis has been shifted from mobility and convenience to environmental friendliness and public health. However, traditional performance measures are insufficient in addressing such changes. Therefore, there is a need to provide a multifaceted approach to integrate such performance measures. Method This study proposed an assessment framework that integrates four aspects: mobility or network performance, traffic safety, environmental impact, and physical activity. In order to demonstrate its applicability, the proposed framework was employed and tested in Michigan as a case study. Data in individual categories were collected and analyzed. Delay costs, metabolic equivalent (MET), the number of crashes, and concentration of O3 were used as performance measures representing network performance, physical activity, safety, and environmental impact, respectively. Result Using the multifaceted assessment framework, this study compared performance measures by region (Bay Region, Grand Region, Metro Region, North Region, Southwest Region, and University Region) in Michigan. The result with data in 2015 showed that the Metro region, the most populated area in Michigan, was weaker than other reasons with respect to network performance, traffic safety and physical activity. By producing a multidimensional spider diagram, the strengths and weaknesses of each region's transportation performance results were analyzed and visualized. The results demonstrated its capacity to present multifaceted performance measures and their relative performances among regions. Conclusion This study proposed an integrated approach and presented a case example using the observed data in Michigan. The method was useful researchers understanding transportation performances in a comprehensive manner integrating multifaceted measures. The approach is expected to help decision-makers in developing traffic management strategies by allowing analyses of traffic-related public health strategies based on traffic exposure measures. In order to examine the effects of alternative strategies, performance measures could be derived from activity-based travel demand models or simulations. Simulation-based assessments can serve as a decision supporting tool for evaluating and/or selecting various treatment options for public health strategies prior to actual implementation. The application of activity-based travel demand models makes it possible to quantify various measures in the proposed approach by providing the effects of alternative strategies.

A Generative Model of Urban Activities from Cellular Data

Activity-based travel demand models are becoming essential tools used in transportation planning and regional development scenario evaluation. They describe travel itineraries of individual travelers, namely, what activities they are participating in, when they perform these activities, and how they choose to travel to the activity locales. However, data collection for activity-based models is performed through travel surveys that are infrequent, expensive, and reflect the changes in transportation with significant delays. Thanks to the ubiquitous cell phone data, we see an opportunity to substantially complement these surveys with data extracted from network carrier mobile phone usage logs, such as call detail records (CDRs). In this paper, we develop input–output hidden Markov models to infer travelers’ activity patterns from CDRs. We apply the model to the data collected by a major network carrier serving millions of users in the San Francisco Bay Area. Our approach delivers an end-to-end actionable solution to the practitioners in the form of a modular and interpretable activity-based travel demand model. It is experimentally validated with three independent data sources: aggregated statistics from travel surveys, a set of collected ground truth activities, and the results of a traffic micro-simulation informed with the travel plans synthesized from the developed generative model.

Automated, electric, or both? Investigating the effects of transportation and technology scenarios on metropolitan greenhouse gas emissions

The anticipated impacts of automated transportation are numerous. Of importance are the effects on energy consumption and greenhouse gas (GHG) emissions. Can automation promote low carbon transportation? Another question often raised relates to the appropriateness of current tools. Will existing tools become irrelevant in the face of the disruptive changes or can we extend the capabilities of our models to broadly capture the effects of automated transportation? This study presents an approach to estimate fuel-cycle GHG emissions, integrated within an activity-based travel demand model for the Greater Toronto and Hamilton Area. The model chain is used to evaluate different shares of automated vehicles (AV), and the effects of electrification of AVs. Daily operating GHG emissions were estimated at 29,000 t in CO2eq with 96% attributed to private vehicles and 4% to transit. While sharing a minor portion of emissions, the public transit system carries 32% of daily passenger kilometers traveled. When accounting for fuel-cycle emissions, the daily total was estimated to be over 36,000 t for private vehicles. With the introduction of AVs, higher vehicle kilometers travelled (3.6%–5.4%) and GHG emissions (2.5%) are expected. However, electrification of AVs can reduce regional GHG emissions (5%), and emission intensities of all vehicles (11%).

A time-use activity-pattern recognition model for activity-based travel demand modeling

This study develops a new comprehensive pattern recognition modeling framework that leverages activity data to derive clusters of homogeneous daily activity patterns, for use in activity-based travel demand modeling. The pattern recognition model is applied to time use data from the large Halifax STAR household travel diary survey. Several machine learning techniques not previously employed in travel behavior analysis are used within the pattern recognition modeling framework. Pattern complexity of activity sequences in the dataset was recognized using the FCM algorithm, and resulted in identification of twelve unique clusters of homogeneous daily activity patterns. We then analysed inter-dependencies in each identified cluster and characterized the cluster memberships through their socio-demographic attributes using the CART classifier. Based on the socio-demographic characteristics of individuals we were able to correctly identify which cluster individuals belonged to, and also predict various information related to their activities, such as start time, duration, travel distance, and travel mode, for use in activity-based travel demand modeling. To execute the pattern recognition model, the 24-h activity patterns are split into 288 three dimensional 5 min intervals. Each interval includes information on activity types, duration, start time, location, and travel mode if applicable. Results from aggregated statistical evaluation and Kolmogorov–Smirnov tests indicate that there is heterogeneous diversity among identified clusters in terms of temporal distribution, and substantial differences in a variety of socio-demographic variables. The homogeneous clusters identified in this study may be used to more accurately predict the scheduling behavior of specific population groups in activity-based modeling, and hence to improve prediction of the times and locations of their travel demands. Finally, the results of this study are expected to be implemented within the activity-based travel demand model, Scheduler for Activities, Locations, and Travel (SALT).

Influencing Mechanism Analysis of Holiday Activity–Travel Patterns on Transportation Energy Consumption and Emissions in China

Energy shortage and atmospheric pollution problems are getting more serious in China, and transportation is the main source of energy consumption, pollutants, and carbon emissions. This study combined the activity-based analysis method with emission models, and investigated the influence mechanism of people’s activity travel scheduling on transportation energy consumption and emissions on holidays. Based on the holiday travel behavior survey data, the multinomial logistic regression model was first applied to explore the decision mechanisms of individual travel-mode choices in holidays. Next, the emission model was integrated with an activity-based travel demand model to calculate and compare transportation energy consumption and emissions under different policy scenarios. The results showed that socio-demographic characteristics had significant effects on holiday activity–travel patterns, and combined mode chains had a larger number of activity points than single mode chains. With an increase in the trip time of cars, and decrease of travel distance and the number of activity points, transportation energy consumption and emissions could be reduced greatly with an adjustment of holiday activity–travel patterns. The reduced portion is mainly attracted by slow traffic and public transport. However, the effects of a single policy strategy are very limited, thus portfolio policies need to be considered by policy makers.

Incorporating in-home activities in ADAPTS activity-based framework: A sequential conditional probability approach

In travel demand modelling literature, minor attention has been given to modelling and specifically micro-simulating the in-home activities. One reason could be associated with the fact that in-home activities do not explicitly involve trips. Besides, data limitation could be another reason since most of the travel surveys are based on the household travel diary approach and do not encompass in-home activity information. Nonetheless, incorporating people's in-home activities into the activity-based travel demand frameworks and perceiving the associations between the activities people perform in home and out of home could enhance the activity generation, planning, and scheduling procedures. In an effort to analyze such associations, this study uses a sequential conditional probability approach for modelling activity type and duration decisions of in-home activities in conjunction with out-of-home ones. The paper first proposes an algorithm to incorporate in-home activities within the Agent-based Activity Planning and Travel Scheduling (ADAPTS) framework. Then, models regarding generation and planning of in-home activities are presented and described accordingly. The new framework is capable of modelling and simulating in-home activities alongside and in conjunction with out-of-home activities with a special focus on a large set of in-home activity types.

Investigating the applicability of ADAPTS activity-based model in air quality analysis

The growing recognition of negative impacts of vehicular emissions on global warming and public health is one of the main factors that led transportation policymakers to devise congestion/pollution mitigation strategies. However, in order to attain the best perception of potential results of implementing these policies, it would be essential to understand the response of people in terms of changing their travel behaviors. Activity-based travel demand models provide comprehensive platforms to simulate individuals’ travel behavior at very fine scale and examine the impact of various policies on population. This paper explores the applicability of activity-based models in evaluating the small- and large-scale environmental consequences of transportation policies. To provide a holistic and policy sensitive platform for estimating mobile source emissions, we integrate an activity-based travel demand model with a mesoscopic dynamic traffic assignment (DTA) module and a microscopic emission simulator. Accordingly, the ADAPTS model is extended with an agent-based DTA component, which allows for dynamic interaction between the demand model and network simulator. Then, the US EPA’s mobile source emission model, MOVES, is leveraged to estimate emissions produced by passenger cars. The main challenge in developing such an integrated framework is compatibility of the data among its different components. The performance of this framework is analyzed through two county- and project-level case studies in Chicago metropolitan area. In the county-level analysis, the base-year replication results show a drastic rise in emissions during peak periods, that is, around 14.8% and 27.2% of total daily NOx emission are produced during morning and evening rush hours, respectively. Further, as the project-level analysis, a capacity improvement in a critical ramp in the network is implemented in the DTA model. The simulation results show 24% reduction in VHT and 2.3% increase in VMT in 10-h simulation period in that link.

Transport habits of travelers using new energy type modes: A random parameters hazard-based approach of travel distance

With environmental concerns on the rise, identifying the factors that affect travel demand for energy efficient and environmentally friendly means of transportation becomes critical. It is, thus, beneficial to identify the characteristics of new energy type – low CO2 emissions – transport habits. Travel time is one such characteristic that has been extensively studied within the context of activity-based travel demand modeling. However, when travelers choose a new energy type with low CO2 emissions transport mode, it is likely that travel time may be playing a secondary role in their mode choice criteria as opposed to travel distance, especially in highly congested urban areas. Within this context, the origin-destination distance of a trip has the potential to reveal interesting traveler preferences and trip patterns. This paper presents an exploratory empirical analysis of origin-destination distance for such new energy type mode transport habits, in the Metropolitan area of Athens, Greece. Using hazard-based econometric modeling and random parameters to account for the longitudinal nature of trip distance and unobserved heterogeneity, respectively, trip distance is found to be affected by a number of influential factors, such as traveler socio-economic and demographic characteristics, trip purpose, trip time, trip frequency, and time of day for the trip.

An analysis of activity timing and mode choice behavior for fixed time workers

Generally, the activity based travel demand modeling was resulted from various professions of individuals such as students, workers and non-workers. The model cannot properly represent travel behavior because their characteristics of activities timing and time allocation in a day significantly different. The purpose of this paper is to analyze travel behavior of out of home non-work activities of working groups who have fixed time working hours within a day in location study Palembang City, Indonesia. The effort to explains the travel behavior is conducted through developing activity timing and mode choice model for out of home non work activities. The activity timing and mode choice model are developed as multinomial logit model by adding the utility function of time allocation for non-work activities. There are 9 alternatives in the model in which the alternatives are the combination of 3 activity time schedule and 3 alternative modes. The result of the model significantly reveals that the characteristic of working individuals in deciding certain modes are not only caused by the mode attributes, such as travel time and cost, but also by the type and time allocation for non-working activities related to that travel.

Activity Duration Analysis: Discrete-Time Approach for Recurrent Events, Competing Risks, and Multiple States

Most existing activity-based travel demand models are typically implemented on a tour-based microsimulation framework. Two approaches are available to describe people’s daily decision-making processes on activity and travel in the activity-based model framework: rubber banding and growing scheduling. The growing-scheduling approach requires a series of linked dynamic discrete choices of activity episodes, locations, and travel modes to build incrementally an entire day’s activity-travel patterns for individuals in households. To incorporate explicitly time dependencies into activity episode choice and duration, a hazard-based duration model in a discrete-time framework is introduced; this framework is essentially the same as a discrete choice model with temporal dummies as the simplest form of dynamic discrete choice models. In the application of the discrete–time duration model to activity duration analysis, complex situations of activity-travel data are considered; these include multiple states of origin activity, competing risks of destination activity, and a multilevel structure for recurrent activity episodes within individuals. In addition, a circular, or periodic, variable is introduced as a combination of sine and cosine to model time-of-day effects.

Identification of Representative Patterns of Time Use Activity Through Fuzzy C-Means Clustering

Analysis of the time use activity patterns of urbanites will contribute greatly to the modeling of urban transportation demands by linking activity generation and activity scheduling modules in the overall activity-based modeling framework. This paper develops a framework for novel pattern recognition modeling to identify groups of individuals with homogeneous daily activity patterns. The framework consists of four modules: initialization of the total cluster number and cluster centroids, identification of individuals with homogeneous activity patterns and grouping of them into clusters, identification of sets of representative activity patterns, and exploration of interdependencies among the attributes in each identified cluster. Numerous new machine-learning techniques, such as the fuzzy C-means clustering algorithm and the classification and regression tree classifier, are employed in the process of pattern recognition. The 24-h activity patterns are split into 288 intervals of 5-min duration. Each interval includes information on activity types, duration, start time, location, and travel mode, if applicable. Aggregated statistical evaluation and Kolmogorov–Smirnov tests are performed to determine statistical significance of clustered data. Results show a heterogeneous diversity in eight identified clusters in relation to temporal distribution and significant differences in a variety of sociodemographic variables. The insights gained from this study include important information on activities—such as activity type, start time, duration, location, and travel distance—that are essential for the scheduling phase of the activity-based model. Finally, the results of this paper are expected to be implemented within the activity-based travel demand model for Halifax, Nova Scotia.

Dynamics of Activity Time-of-Day Choice

This study explored travelers’ decision behavior in selecting activity start times. The study examined the problem in the context of the Agent-based Dynamic Activity Planning and Travel Simulation (ADAPTS) activity-based travel demand model for the Chicago, Illinois, metropolitan area. A unique feature of the ADAPTS framework is its consideration of planning horizons for various activity attributes. Naturally, the various attributes of an activity—such as start time, duration, location, party involvement, and mode of travel—can be planned in different time horizons. An attribute that is planned affects the choice of other activity attributes. Therefore, developing a true behavioral time-of-day choice model would not be possible unless the planning order of activity attributes and the dynamics of travelers’ decision-making processes are taken into account. Similarly, it can be argued that there should be fundamental differences in the time-of-day decision process when other attributes of the activity are not yet planned but are to be decided at a later time. The presented time-of-day model aims to capture the dynamics of this decision process by considering the planning time horizons of other attributes of the activity, as well as the outcomes of the decisions. The study adopted the discrete choice approach to model activity timing decisions and a hybrid utility maximization and developed a regret minimization model to account for the heterogeneity of decision rules across choice variables. Analysis of the estimation results and parameter elasticities indicates that higher expected travel time, variations in travel time, and schedule occupancy rates for different time choices can significantly increase the regret value of the corresponding choice and therefore affect the time-of-day choice.

Looking beyond the mean for equity analysis: Examining distributional impacts of transportation improvements

Activity-based travel demand models can be useful tools for understanding the individual level equity impacts of transportation plans, because of their ability to generate disaggregate transportation measures. However, these capabilities have yet to be fully explored in public practice. In this paper we first discuss a general framework for performing transportation equity analysis using activity-based travel demand models, distributional comparisons, and incorporating equity standards. In addition, we demonstrate the advantages of distributional comparisons, relative to average measures. This demonstration uses the 2000 Bay Area Travel Survey and (activity-based) mode choice model. The findings show that distributional comparisons are capable of clearly revealing the winners and losers that result from transportation improvements, in comparison with average measures. The use of these results will likely result in different conclusions on transportation investments.