Travel Demand Forecasting and Model Application

APA Florida Chapter, Broward Section


Friday, March 15, 2019
9 a.m. - noon CDT

Fort Lauderdale, FL, United States

CM | 2.25

Add to My Log


BAPA continues its partnership with South East Florida FSUTMS Users Group of FDOT. This event focuses on new ways of aquiring and using multiple forms of taffic data.

The first presentation is titled “Cell Phone Location Data for Travel Behavior Analysis”, 

The transportation community has a great interest in how different types of location data can be analyzed to infer travel patterns in a region to support planning and modeling applications which are traditionally based on survey data and regional models.  A key component of this discussion is the comparison and contrast between new forms of location data with traditional household surveys, Census based products such as the CTPP, and regional model outputs.  NCHRP Report 868: Cell Phone Location Data for Travel Behavior Analysis, presents guidelines for transportation planners and travel modelers on how to:

(1) evaluate the extent to which cell phone location data accurately depict travel; 

(2) identify whether and how these data can be used to improve our understanding of travel characteristics and our ability to model travel patterns and behavior more effectively; and 

(3) support practitioners’ evaluation of the strengths and weaknesses of anonymized “call detail record” (CDR) locations from cell phone data.

This guidebook is intended for transportation practitioners and agency staff interested in new methods of capturing travel data from cell phones.  This is an emerging field of interest which is subject to complexities linked to acquiring data and applying these data while maintaining privacy in a complex legal and practical framework.  

The emergence of these data constitutes a significant opportunity for change in the travel modeling community, with access to detail and volume not previously available.  Therefore, a better understanding of the strengths and weaknesses of these data is an important step in this direction.  

With the emergence of large amounts of data, research is needed to explore and evaluate methods used for processing cell phone location data to generate travel behavior information and provide guidelines for the use of the information by transportation planning practitioners.  A key issue, which does not have a clear or straightforward answer, is which of these sources constitutes “ground truth.”  

In this study we compare CDR-derived results with household surveys, Census estimates of commute travel, and well understood regional model outputs. However, we need to recognize the different nature of each data source and the fact that all data sources reflect a sample of observations that has strengths, weaknesses, assumptions, and errors embedded in it.  A related question includes the assumptions made and inferences drawn when analyzing each data source.  Weaknesses in each analysis approach reflect different assumptions used about how travel characteristics are inferred from underlying data.

The study used travel in the Boston region as a case study to compare and contrast traditional travel survey data, regional models, and Census data with cell phone derived data describing regional travel.  The report summarizes these comparisons and also provides a “how to” guide for the evaluation of similar sources of location data both today and in the future.


The second presentation is titled “Traffic Forecasting Tool for Indiana Department of Transportation”, 

 This presentation describes an R-based Traffic Forecasting Tool (TFT) under development for the Indiana Department of Transportation (INDOT). As a post-processor of the Indiana Statewide Travel Demand Model (ISTDM), the TFT incorporates traffic forecasts using both ISTDM data and the statewide historical traffic counts. The TFT provides various traffic forecasting methods that are commonly used by INDOT, such as linear, regression and growth factor models. The TFT also incorporates refined methods (i.e., Indiana Traffic Growth Profile, and NCHRP 255) to enhance traffic forecasting results. This tool is suitable for long-term traffic forecasting at project level for all Indiana state jurisdictional roadways. It provides effective connectivity between INDOT’s traffic count database and the ISTDM.  

The INDOT’s existing TFT was developed based on TransCAD/GISDK, while the newer R-based version builds upon the open-source R programming language along with background ISTDM and traffic count datasets in the common formats of CSV and shapefiles to generate desired outputs. The R-based TFT is an evolved and more promising version, as it allows for a better output visualization as well as an easier on-line access and application of the TFT tool via personal computers, tablets or smart phones, whenever and wherever, even without the TransCAD software.


The third presentation is titled “A Systematic Traffic Count Balance Method in Network Levels”, 

 Currently, traffic balance works involve tremendous manual efforts and are difficult to be implemented in network levels. This presentation proposes a systematic approach for traffic balance in corridor/network levels. With the assumption on the correctness for most of the collected traffic counts, a mathematic optimization model was developed to minimize the weighted sum square of the traffic count adjustment with constrains on the equilibrium of traffic counts (consistent counts) in basic segments and in multiple time periods.  The basic segment can be a node in the network with complete observed count for all its connected legs.  The basic segment can also be a minimal mainline-to-mainline part of a freeway. The weight factors in the model stand for the confidence the user wants to put on an individual traffic count.  The model can be easily expanded from an isolated corridor/node level to a whole network level with the consistent balance results generated simultaneously for multiple facilities and multiple time periods.  In addition, a CUBE based application was also developed to prepare the inputs for the traffic balance model. Based on the depth First Search (DFS) algorithm, the application can be used to extract traffic counts in a proper topologic order from a corridor or from multiple connected nodes. Finally the presentation demonstrates the implementation of the propose method on the D5 CFRPM network.


The fourth presentation is titled “From Manual to Automation – TMC Visualization and Balancing Tool”,  

The TMC Visualization and Balancing tool is meant to streamline and automate some of the efforts that it takes for an analyst to develop and report turning movement forecasts. It is also expected to help automate QA/QC of the data collected as part of design traffic studies. Currently, it performs the following: (1) outputs plots (interactive and static) of unbalanced flows, (2) automatically balances turning movements, and (3) output plots (interactive and static) of balanced flows. Future extension possibilities include steps to automatically generate and plot TMC forecasts directly from the raw TMC counts. The tool is developed in R and allows users to interact and customize for different corridors/projects using an input Excel file.

Resource URL


Kimon E Proussaloglou, Ph.D.

Dr, Proussaloglou is an Executive Vice President with Cambridge Systematics. He leads its demand modeling, data analytics and market research practice and is the director of the Chicago office. He received both a Doctorate and a M.Sc. in Civil Engineering from Northwestern University and a B.Sc. from Aristotelian ... Read More

Akbar Bakhshi, EIT

Mr. Bakhshi is a modeler in Corradino’s transportation modeling practice with more than 5 years of experience in travel demand modeling development, application, transportation planning and engineering. His recent work has been focused on regional & statewide travel demand modeling for MPOs and State DOTs in Indiana, Michigan, Kentucky and ... Read More

Hongbo Chi, Ph.D. PE

Dr. Chi is a transportation modeler with AECOM in Miami, Florida. He has 8 years of transportation modeling experience, 5 years with AECOM and 3 years with Citilabs. He is a licensed professional engineer in Florida. He has extensive experiences on traffic and transit model development, calibration and validation. He ... Read More

Ashutosh Kumar

Ashutosh (Ashu) has nearly 15 years of experience in multi-modal corridor planning studies, data driven planning techniques, travel demand model development and forecasting, traffic projections, transit ridership estimates and New Starts analyses. In addition to his extensive use of a variety of travel demand forecasting software packages, Ashu continue to ... Read More

Contact Info

Marilyn Mammano,