Presentation is loading. Please wait.

Presentation is loading. Please wait.

Making advanced travel forecasting models affordable through model transferability 14th TRB Conference on Transportation Planning Applications May 5-9,

Published byDomenic Webster Modified over 9 years ago

Similar presentations

Presentation on theme: "Making advanced travel forecasting models affordable through model transferability 14th TRB Conference on Transportation Planning Applications May 5-9,"— Presentation transcript:

1 Making advanced travel forecasting models affordable through model transferability 14th TRB Conference on Transportation Planning Applications May 5-9, 2013, Columbus, Ohio John L Bowman, Mark Bradley, Joe Castiglione, Supin Yoder

2 Acknowledgments Sponsored by FHWA under the STEP program Data provided by California DOT Fehr & Peers Florida DOT Fresno County COG Sacramento Area COG San Joaquin County COG San Diego Association of Governments TRB Planning Apps Conf May 2013Bowman, et al2

3 TRB Planning Apps Conf May 2013Bowman, et al3 Outline Introduction Transferability testing methods Results

4 TRB Planning Apps Conf May 2013Bowman, et al4 Objective Empirically test and demonstrate the transferability of activity-based (AB) models between regions Why? Reduce AB development costs large household survey estimating entirely new models

5 Six regions in study TRB Planning Apps Conf May 2013Bowman, et al5 Sacramento San Diego Northern San Joaquin Valley Jacksonville Tampa Fresno

6 Activity-Based Models: 1993-2012John L Bowman, Ph.D. (www.JBowman.net)6 AB Model Framework

7 Fifteen tested model components TRB Planning Apps Conf May 2013Bowman, et al7 Model TypeNumber of coefficients Usual work location48 Auto ownership24 Person-day tour generation126 Exact number of tours86 Work tour time of day69 Work tour mode (detailed LOS)58 Work tour mode (combined LOS)31 Work-based subtour generation14 School tour mode32 Other tour destination62 Other tour time of day86 Other tour mode41 Intermediate stop generation100 Intermediate stop location66 Trip time of day45 Total888

8 Seven untested model components TRB Planning Apps Conf May 2013Bowman, et al8 Model Type Usual school location Work tour destination Escort tour mode Work-based subtour mode School tour time of day Work-based subtour time of day Trip mode

9 Eleven variable types TRB Planning Apps Conf May 2013Bowman, et al9 Variable Type Number of coefficients A-constant192 P-person184 H-household149 D-day pattern76 T-tour/trip199 I-impedance110 U-land use99 W-time window45 C-logsum24 G-size variable35 L-log size multiplier3 Total1116

10 Transferability testing: two approaches Application-Based Apply model system developed for another region Compare predictions to observed aggregate outcomes Estimation-Based Estimate coefficients for both regions Compare them for statistical differences TRB Planning Apps Conf May 2013Bowman, et al10

11 Strengths of the estimation-based approach Explicit statistical tests Can address a wide variety of hypotheses Can test transferability of specific variable types and model components TRB Planning Apps Conf May 2013Bowman, et al11

12 Data issues Data problems can confound transferability test results Inconsistent data Small samples TRB Planning Apps Conf May 2013Bowman, et al12

13 Estimability questions What estimation sample size is adequate? How does combining samples improve estimability? Which models are more estimable at the regional level? TRB Planning Apps Conf May 2013Bowman, et al13

14 Transferability Hypothesis 1 Variables that apply to population segments defined by characteristics of individuals and/or their situational context (i.e. segment-specific variables) will tend to be more transferable than variables that are more generic and apply to all individuals. TRB Planning Apps Conf May 2013Bowman, et al14

15 Transferability Hypothesis 2 Variables that are segment-specific will tend to be more transferable than alternative-specific constants. TRB Planning Apps Conf May 2013Bowman, et al15

16 Transferability Hypothesis 3 Models that deal with social organization (activity generation and scheduling) will be more transferable than models that deal mainly with spatial organization (mode choice and location choice) TRB Planning Apps Conf May 2013Bowman, et al16

17 Transferability Hypothesis 4 Models for different regions within the same state will tend to be more transferable than models for regions in different parts of the country TRB Planning Apps Conf May 2013Bowman, et al17

18 TRB Planning Apps Conf May 2013Bowman, et al18 Outline Introduction Transferability testing methods Results

19 Testing method overview Prepare data Estimate separate models Estimate comparison models Tabulate and analyze results TRB Planning Apps Conf May 2013Bowman, et al19

20 Data preparation overview TRB Planning Apps Conf May 2013Bowman, et al20 For each region:

21 NHTS Sample Sizes RegionNumber of Households Fresno380 Northern San Joaquin Valley660 Sacramento1,310 San Diego6,000 California Total8,350 Jacksonville1,050 Tampa2,500 Florida Total 3,550 Two-state total 11,900 TRB Planning Apps Conf May 2013Bowman, et al21

22 Testing method overview Prepare data Estimate separate models Estimate comparison models Tabulate and analyze results TRB Planning Apps Conf May 2013Bowman, et al22

23 Estimating separate models Generate base model specs Estimate 90 separate models (15 models x 6 regions) Constrain inestimable coefficients TRB Planning Apps Conf May 2013Bowman, et al23

24 Testing method overview Prepare data Estimate separate models Estimate comparison models Tabulate and analyze results TRB Planning Apps Conf May 2013Bowman, et al24

25 Estimating comparison models For each of the 15 models : Combine data for all regions Estimate 36 versions of each model 12 base model versions 24 difference model versions TRB Planning Apps Conf May 2013Bowman, et al25

26 Utility functions TRB Planning Apps Conf May 2013Bowman, et al26

27 Testing Method Overview Prepare data Estimate separate models Estimate comparison models Tabulate and analyze results TRB Planning Apps Conf May 2013Bowman, et al27

28 TRB Planning Apps Conf May 2013Bowman, et al28 Outline Introduction Transferability testing methods Results Hypotheses tested Most important conclusion

29 Transferability hypotheses (Hypothesis 1) Segment-specific variables will tend to be more transferable than variables that are more generic and apply to all individuals. (accepted) (Hypothesis 2) Variables that are segment- specific will tend to be more transferable than alternative-specific constants. (rejected) TRB Planning Apps Conf May 2013Bowman, et al29

30 TRB Planning Apps Conf May 2013Bowman, et al30

31 Transferability hypotheses (Hypothesis 3) Models that deal with social organization (activity generation and scheduling) will be more transferable than models that deal mainly with spatial organization (mode choice and location choice) (accepted) TRB Planning Apps Conf May 2013Bowman, et al31

32 1 TRB Planning Apps Conf May 2013Bowman, et al32

33 Transferability hypotheses (Hypothesis 4) Models for different regions within the same state will tend to be more transferable than models for regions in different parts of the country California—accepted (weakly) Florida—rejected, Jacksonville more transferable with California than with Tampa TRB Planning Apps Conf May 2013Bowman, et al33

34 TRB Planning Apps Conf May 2013Bowman, et al34 Outline Introduction Transferability testing methods Results Hypotheses tested Most important conclusions

35 Most Important Conclusions evidence of broad comparability among all the regions, with one region, Tampa, standing out as less comparable than the others sample sizes of 6,000 households or more provide much better information for estimating coefficients than samples of 2,500 or less. Better to transfer models based on large sample from comparable region than to estimate new models using a much smaller local sample TRB Planning Apps Conf May 2013Bowman, et al35

36 Comparability—2 state TRB Planning Apps Conf May 2013Bowman, et al36

37 Comparability within state TRB Planning Apps Conf May 2013Bowman, et al37

38 Most Important Conclusions evidence of broad comparability among all the regions, with one region, Tampa, standing out as less comparable than the others sample sizes of 6,000 households or more provide much better information for estimating coefficients than samples of 2,500 or less. Better to transfer models based on large sample from comparable region than to estimate new models using a much smaller local sample TRB Planning Apps Conf May 2013Bowman, et al38

39 Need for large samples TRB Planning Apps Conf May 2013Bowman, et al39

40 Most Important Conclusions evidence of broad comparability among all the regions, with one region, Tampa, standing out as less comparable than the others sample sizes of 6,000 households or more provide much better information for estimating coefficients than samples of 2,500 or less. Better to transfer models based on large sample from comparable region than to estimate new models using a much smaller local sample TRB Planning Apps Conf May 2013Bowman, et al40

41 TRB Planning Apps Conf May 2013Bowman, et al41

42 12 Base model versions 6 region-specific versions 2 2-state versions With region-specific ASCs Without region-specific ASCs 2 FL versions (with & without region ASCs) 2 CA versions (with & without region ASCs) TRB Planning Apps Conf May 2013Bowman, et al42

43 24 Difference model versions Four versions for each region each with difference variables relative to a base version 2 state base 2 state base with region-specific ASCs 1 state base 1 state base with region-specific ASCs TRB Planning Apps Conf May 2013Bowman, et al43

44 TRB Planning Apps Conf May 2013Bowman, et al44

45 TRB Planning Apps Conf May 2013Bowman, et al45

Download ppt "Making advanced travel forecasting models affordable through model transferability 14th TRB Conference on Transportation Planning Applications May 5-9,"

Similar presentations

GIS and Transportation Planning

GIS and Transportation Planning
Denver’s Activity-Based Model Project: Status Report Erik Sabina, P.E., DRCOG Thomas Rossi, Principal, Cambridge Systematics TRB Transportation Planning.

Denver’s Activity-Based Model Project: Status Report Erik Sabina, P.E., DRCOG Thomas Rossi, Principal, Cambridge Systematics TRB Transportation Planning.
NCHRP 08-78 Renaissance Planning Group Rich Kuzmyak Chris Sinclair Alex Bell TRB National Transportation Planning Applications Conference May 6, 2013 Columbus,

NCHRP Renaissance Planning Group Rich Kuzmyak Chris Sinclair Alex Bell TRB National Transportation Planning Applications Conference May 6, 2013 Columbus,
Transportation leadership you can trust. presented to TRB Planning Applications Conference presented by Elizabeth Sall Maren Outwater Cambridge Systematics,

Transportation leadership you can trust. presented to TRB Planning Applications Conference presented by Elizabeth Sall Maren Outwater Cambridge Systematics,
The transition to activity-based models in the U.S. Mark Bradley Bradley Research & Consulting Santa Barbara, CA.

The transition to activity-based models in the U.S. Mark Bradley Bradley Research & Consulting Santa Barbara, CA.
FOCUS MODEL OVERVIEW CLASS TWO Denver Regional Council of Governments June 30, 2011.

FOCUS MODEL OVERVIEW CLASS TWO Denver Regional Council of Governments June 30, 2011.
Norman Washington Garrick CE 2710 Spring 2014 Lecture 07

Norman Washington Garrick CE 2710 Spring 2014 Lecture 07
Presented to presented by Cambridge Systematics, Inc. Transportation leadership you can trust. Comparison of Activity-Based Model Parameters Between Two.

Presented to presented by Cambridge Systematics, Inc. Transportation leadership you can trust. Comparison of Activity-Based Model Parameters Between Two.
FOCUS MODEL OVERVIEW CLASS THREE Denver Regional Council of Governments July 7, 2011.

FOCUS MODEL OVERVIEW CLASS THREE Denver Regional Council of Governments July 7, 2011.
Time of day choice models The “weakest link” in our current methods(?) Change the use of network models… Run static assignments for more periods of the.

Time of day choice models The “weakest link” in our current methods(?) Change the use of network models… Run static assignments for more periods of the.
Development of a New Commercial Vehicle Travel Model for Triangle Region 14 th TRB Planning Applications Conference, Columbus, Ohio May 7, 2013 Bing Mei.

Development of a New Commercial Vehicle Travel Model for Triangle Region 14 th TRB Planning Applications Conference, Columbus, Ohio May 7, 2013 Bing Mei.
Opportunities & Challenges Using Passively Collected Data In Travel Demand Modeling 15 th TRB Transportation Planning Applications Conference Atlantic.

Opportunities & Challenges Using Passively Collected Data In Travel Demand Modeling 15 th TRB Transportation Planning Applications Conference Atlantic.
A National County-Level Long Distance Travel Model Mike Chaney, AICP Tian Huang, PE, AICP, PTOE Binbin Chen, AICP 15 th TRB National Transportation Planning.

A National County-Level Long Distance Travel Model Mike Chaney, AICP Tian Huang, PE, AICP, PTOE Binbin Chen, AICP 15 th TRB National Transportation Planning.
Tour-based and Supply Chain Modeling for Freight in Chicago May 9, 2013 Prepared for: TRB Planning Applications Conference Prepared by: Maren Outwater.

Tour-based and Supply Chain Modeling for Freight in Chicago May 9, 2013 Prepared for: TRB Planning Applications Conference Prepared by: Maren Outwater.
San Francisco Bay Area Activity-Based Models Specification & Training Study Chuck Purvis Metropolitan Transportation Commission Oakland, California Workshop.

San Francisco Bay Area Activity-Based Models Specification & Training Study Chuck Purvis Metropolitan Transportation Commission Oakland, California Workshop.
Overview of the Research Process in Economics Researchers in Economics, as a social science, use a version of the scientific method. The scientific method.

Overview of the Research Process in Economics Researchers in Economics, as a social science, use a version of the scientific method. The scientific method.
Model Task Force Meeting November 29, 2007 Activity-based Modeling from an Academic Perspective Transportation Research Center (TRC) Dept. of Civil & Coastal.

Model Task Force Meeting November 29, 2007 Activity-based Modeling from an Academic Perspective Transportation Research Center (TRC) Dept. of Civil & Coastal.
Problem Statement and Motivation Key Achievements and Future Goals Technical Approach Kouros Mohammadian, PhD and Yongping Zhang (PhD Candidate), CME,

Problem Statement and Motivation Key Achievements and Future Goals Technical Approach Kouros Mohammadian, PhD and Yongping Zhang (PhD Candidate), CME,
Transportation leadership you can trust. presented to Transportation Planning Applications Committee (ADB50) presented by Sarah Sun Federal Highway Administration.

Transportation leadership you can trust. presented to Transportation Planning Applications Committee (ADB50) presented by Sarah Sun Federal Highway Administration.
Presentation Outline Project Purpose and ApproachProject Purpose and Approach Review of Existing Modeling Practice in CaliforniaReview of Existing Modeling.

Presentation Outline Project Purpose and ApproachProject Purpose and Approach Review of Existing Modeling Practice in CaliforniaReview of Existing Modeling.

Similar presentations

Ads by Google