What are the Different Demand Forecasting Models?
Travel demand refers to the expected number of passengers or vehicles that will travel on a given part of a transportation system, taking into account factors such as land use, socio-economics and the environment. Travel demand forecasts are crucial for predicting future or changing vehicular traffic on transportation systems. Travel demand forecasting methods can range from simple estimates to complex computerized processes, depending on project constraints such as data availability and funding (Garber & Hoel, 1999).
Urban travel demand is influenced by three main factors: land use, socio-economic characteristics of the population, and the quality and accessibility of transport services. Different types of land uses, such as residential, commercial or office areas, can generate traffic. Apart from land use, socio-economic factors such as income and lifestyle influence how people use transportation resources. The quality and accessibility of transportation services, including factors such as cost, ease of use and safety, are also examples of factors that influence people's travel decisions (Garber & Hoel, 1999).
The traditional 4-step model structure is shown in Figure 1. Trip-based models, such as the traditional 4-step model, were among the first demand forecasting models to be used.
Tour-based models emerged as an important development in the late 1970s and 1980s. The tour-based models shown in Figure 2 represent the cycle of trips within a tour. While this model takes into account temporal and spatial constraints, it does not bind multiple trips within the same time period.
The activity-based model structure is shown in Figure 3. Activity-based models are based on the willingness to allocate travel demand across activities. Activity-based models aim to predict the activity and travel schedules of individuals over a given time period, taking into account temporal and spatial constraints (Omer, Kim, Sasaki & Nishii, 2010).
Different demand forecasting models are used to answer questions about why, how often, where and how people travel. In the literature, there are different types of modeling such as Standard 4-Step Model, Tour-Based Model, Activity-Based Model.
Once the study area and issues have been identified before starting the technical aspects of travel forecasting, it is important to divide the study area into Transportation/Traffic Analysis Zones (TAZ) to cover the policy impacts identified. These zones can be of different sizes and can also be grouped into larger zones for specific analyses (Garber & Hoel, 1999).
The 4-step model is used as the main tool for forecasting the future demand and efficiency of the transportation system (McNally, 2007). Each step in the 4-step demand model aims to address a different question:
How much total travel demand is there?
Where are the potential travel destinations located?
Which travel modes are suitable for these trips?
Which routes will be chosen to complete the trips?
In a simple network the demand functions can be estimated directly, but in a complex network a modeling is required for a realistic implementation. Therefore, the 4-step model is formulated as a model with four consecutive steps (Figure 4) (McNally, 2007).
The 4-step model can be summarized as in Figure 5 below:
Trip Generation is the first step of the 4-step model. In the trip generation step, the number of trips in a certain time period and in a certain area is expressed. In this step, it is aimed to obtain the travel trend and trip frequency within the network. The trips that occur can be represented as trip ends, production and attraction points (McNally, 2007). The 4-step model is the main method for demand forecasting and is based on land use and travel characteristics (Garber & Hoel, 1999).
The second step is known as Trip Distribution and refers to how trips occurring in the study area are distributed. In the trip distribution step, distance, duration, mode of transportation, density, attraction areas and other factors are taken into account to estimate the distribution of trips. In short, in the trip distribution step, the generated trips are matched with the distribution of trip attraction points and distributed by considering factors such as time and/or cost, which are travel impedance values (McNally, 2007).
The third step is called Mode Choice. In this step, it is aimed to determine which transportation mode will be preferred by considering the travel tables created in the previous steps.
The last step of the 4-step model is called Network Assignment. It aims to distribute the existing travel tables with the transportation modes added in the third step to the relevant alternative routes (McNally, 2007). Briefly, Route Selection can be defined as the process of determining which route individuals will choose to travel from a given source to another point.
In the tour-based model, tours form the basis of the analysis. A tour can be defined as a closed chain of a series of activities starting and ending at a specific location. For each tour, there are two different destinations, a primary destination and a secondary destination. The primary destination is defined as the place where the most important activity takes place. The secondary destination is the destination of any vehicle journey during the tour. Trips refer to the journeys that occur between the start and the primary destination or secondary destination during the tour (Sener, Ferdous, Bhat & Reeder, 2009).
Figure 6 shows the tour of a parent dropping off their child at school and then going to work. In this tour, the starting point is home and the primary destination is work, while the secondary destination is the child's school. During this tour the parent completes 3 journeys: home to school, school to work and work to home. The tour ends at the starting point, which is home.
There are some key differences between the Trip-Based Model and the Tour-Based Model (Sener, Ferdous, Bhat, & Reeder, 2009). These differences are explained in Table 1.
Table 1 Differences between Trip-Based Model and Tour-Based Model
Activity-Based Models have gained more popularity than the traditional 4-step model and offer improvements to the 4-step model. Both types involve the generation of activities, setting goals, identifying transportation modes and estimating network travel routes. However, activity-based models stand out for their ability to link the activities and travel of individuals and households, explicitly taking into account realistic time and space constraints (Figure 7). This allows for a more accurate representation of how travel conditions affect personal choices. Activity-based models include detailed individual and household characteristics and provide comprehensive performance measures thanks to their detailed person-level approach, in contrast to the region-level focus of most travel-based models (National Academies of Sciences, Engineering, and Medicine, 2014).
Figure 7 Standard Activity-Based Modeling Structure
(Adapted from National Academies of Sciences, Engineering, and Medicine, 2014)
Activity-based travel models are based on the idea that travel demand is driven by people's needs and desires to engage in activities. These models are based on behavioral theories that take into account various constraints on how people make decisions about participating in activities, taking into account where to do the activity, when to do it, and how to get to the destination.
Compared to travel-based models, activity-based models stand out in several ways. Table 2 shows the differences between the trip-based model and the activity-based model. This model comprehensively represents the activity and travel preferences of each individual throughout the day and takes into account the types of activities and their ranking priorities. Activity-based models provide a more realistic representation for assessing the impact of investments, policies or other changes on people's travel preferences.
Table 2 Comparison of Travel-Based Model and Activity-Based Model
Garber N. J. & Hoel L. A. (1999). Traffic and highway engineering (2nd ed. rev. print). PWS Pub. Retrieved from https://users.pfw.edu/sahap/CE450%20Transport%20Policy%20and%20Planning/1.%20Lectures/Books,%20references,%20readings/Chapter%2012%20Forecasting%20Travel%20Demand.pdf
IOWA State University Lecture Notes. (2015). Retrieved from https://www.youtube.com/watch?v=i6JsaC5Hxuk&ab_channel=CE355PrinciplesofTransportationEngineering Accessed on October 9, 2023.
McNally, M.G. (2007), "The Four-Step Model", Hensher, D.A. and Button, K.J. (Ed.) Handbook of Transport Modelling (, Vol. 1), Emerald Group Publishing Limited, Bingley, pp. 35-53. https://doi.org/10.1108/9780857245670-003 Retrieved from https://escholarship.org/uc/item/0r75311t
National Academies of Sciences, Engineering, and Medicine. 2014. Activity-Based Travel Demand Models: A Primer. Washington, DC: The National Academies Press. https://doi.org/10.17226/22357.
Omer, M., Kim, H., Sasaki, K. et al. (2010). A tour-based travel demand model using person trip data and its application to advanced policies. KSCE J Civ Eng 14, 221–230. https://doi.org/10.1007/s12205-010-0221-6 Retrieved from https://www.researchgate.net/publication/227197503_A_Tour-based_Travel_Demand_Model_Using_Person_Trip_Data_and_Its_Application_to_Advanced_Policies
Sener, I. N., Ferdous, N., Bhat, C. R., & Reeder, P. (2009). Tour-based model development for TxDOT: evaluation and transition steps (No. FHWA/TX-10/0-6210-2). University of Texas at Austin. Center for Transportation Research. Retrieved from https://rosap.ntl.bts.gov/view/dot/18156
An urban planner who wants to touch real life with the transportation sector. She is interested in sustainable transport, traffic safety and urban mobility.