Effect on Model Sensitivities of Combining Transferable Data from Separate Home Interview Surveys

Effect on Model Effect on Model Sensitivities of Combining Sensitivities of Combining Transferable Data from Transferable Data from Separate Home Interview Separate Home Interview SurveysSurveys

Presented to the 11Presented to the 11thth Conference on Conference on Transportation Planning ApplicationsTransportation Planning Applications

May 8, 2007May 8, 2007

ByBy

Jonathan Avner, Wilbur Smith AssociatesJonathan Avner, Wilbur Smith AssociatesGregory Giaimo, Ohio Department of Gregory Giaimo, Ohio Department of TransportationTransportation

OutlineOutline

Analysis of Data TransferabilityAnalysis of Data Transferability– Production RatesProduction Rates– Trip Length AnalysisTrip Length Analysis– Time of Day AnalysisTime of Day Analysis

Sensitivity of Models with Sensitivity of Models with Transferred DataTransferred Data

BackgroundBackground

ODOT undertook a household survey ODOT undertook a household survey data collection effort in 2000 to data collection effort in 2000 to support the development of a new support the development of a new generation of travel demand models in generation of travel demand models in the small and medium sized MPOs.the small and medium sized MPOs.

In total, over sixteen thousand In total, over sixteen thousand households were surveyed (MPO and households were surveyed (MPO and non MPO areas) that included more non MPO areas) that included more than 100,000 trip records.than 100,000 trip records.

Survey Household Survey Household LocationsLocations

Data TransferabilityData Transferability

Data TransferabilityData Transferability

Previous research has focused on feasibility of Previous research has focused on feasibility of avoiding surveying by borrowing other data.avoiding surveying by borrowing other data.

This research focused on combining data to This research focused on combining data to obtain improved parameter estimates. obtain improved parameter estimates.

Each area had 1300 to 1900 households Each area had 1300 to 1900 households surveyed and would be getting the same surveyed and would be getting the same model design with calibrated parameters.model design with calibrated parameters.

Considered following model componentsConsidered following model components– Trip Production RatesTrip Production Rates– Trip Distribution (Friction Factor Calibration)Trip Distribution (Friction Factor Calibration)– Time of Day Time of Day

Areas Considered for Areas Considered for CombinationCombination

MPOMPO SmSmallall

LargLargee

Group Group 11

Group Group 22

ToledoToledo XX

LimaLima XX XX

DaytonDayton XX

SpringfieldSpringfield XX XX

AkronAkron XX XX

CantonCanton XX XX

MansfieldMansfield XX XX

SteubenvilSteubenvillele

XX

YoungstowYoungstownn

xx XX

Trip Production Rate Trip Production Rate Analysis - PurposeAnalysis - Purpose Determine whether datasets could be Determine whether datasets could be

combined to create larger estimation datasets combined to create larger estimation datasets for better parameter estimation.for better parameter estimation.

Depending on purpose, trip rates are stratified Depending on purpose, trip rates are stratified by wealth (vehicles / hh), size (hh size, hh by wealth (vehicles / hh), size (hh size, hh workers, etc.) and possibly area type.workers, etc.) and possibly area type.

With combined datasets able to achieve With combined datasets able to achieve minimum number of observations per cell with minimum number of observations per cell with area type stratification (not necessarily area type stratification (not necessarily without).without).

Thus if area type dimension needed, combining Thus if area type dimension needed, combining study area datasets could be necessary.study area datasets could be necessary.

Trip Production Rate Trip Production Rate Analysis – Statistical Analysis – Statistical AnalysisAnalysis The mean trip production rate was compared The mean trip production rate was compared

on a cellular basis for each combination on a cellular basis for each combination (small, large, group 1, group 2).(small, large, group 1, group 2).

ANOVA (analysis of variance) was used since ANOVA (analysis of variance) was used since greater than two samples were being greater than two samples were being considered.considered.

Results were based on looking at F statistic:Results were based on looking at F statistic:– Ratio between the group variability and within Ratio between the group variability and within

group variabilitygroup variability– Value close to 1 Value close to 1 → accept H→ accept Hoo (means are equal) (means are equal)– Value much >1 → reject HValue much >1 → reject Hoo (means are not equal) (means are not equal)

Trip Production Rate Trip Production Rate Analysis – Area TypeAnalysis – Area Type For the larger MPOs, the For the larger MPOs, the

trip rates were trip rates were compared between area compared between area types to determine need types to determine need for this dimension.for this dimension.

Four area types are used Four area types are used in generation: CBD, in generation: CBD, Urban, Suburban, RuralUrban, Suburban, Rural

Average trip rates Average trip rates between area types in between area types in the large MPOs were the large MPOs were tested using ANOVA.tested using ANOVA.

Trip Production Rate Trip Production Rate Analysis – Area TypeAnalysis – Area Type High F statistics indicated High F statistics indicated

difference between difference between average trips between average trips between different area types.different area types.

Unique production rates Unique production rates were calibrated for area were calibrated for area types or combinations of types or combinations of area types when:area types when:– F statistic was large F statistic was large

between area types; andbetween area types; and– Sample size large Sample size large

enough in each cell enough in each cell Households per cell>30Households per cell>30

0

1

2

3

4

5

6

7

8

F Stat

HBW HBSH HBO NHBW NHBO

Trip Production Rate Trip Production Rate Analysis - ResultsAnalysis - Results Chosen combination of Chosen combination of

study area data would be study area data would be applied to all trip purposes applied to all trip purposes in the trip generation in the trip generation modelmodel

Necessary to develop Necessary to develop overall “score” for each overall “score” for each combination, since actual combination, since actual ANOVA at a cellular levelANOVA at a cellular level– Households in each cell of Households in each cell of

combination were added combination were added together if cell had together if cell had significant F statistic significant F statistic (accept H(accept Hoo))

– Results below indicate Results below indicate percentage of households percentage of households that are in cells with that are in cells with similar trip rates.similar trip rates.

0 20 40 60 80 100

Group 2

Group 1

Large

Small

Co

mb

inat

ion

Percent of Households

Average

NHBO

NHBW

HBO

HBSH

HBW

HBHBWW

HBSHBSHH

HBOHBO NHNHWW

NHONHO AvgAvg

SmalSmalll

61%61% 29%29% 50%50% 26%26% 61%61% 45%45%

LargLargee

72%72% 68%68% 70%70% 69%69% 70%70% 70%70%

Grp Grp 11

68%68% 58%58% 57%57% 89%89% 59%59% 66%66%

Grp Grp 22

51%51% 67%67% 60%60% 66%66% 71%71% 63%63%

Trip Production Rate Trip Production Rate Analysis - Analysis - RecommendationsRecommendations

AreaArea CombineCombine

ToledoToledo LargeLarge

LimaLima Group 1Group 1

DaytonDayton

SpringfieldSpringfield Group 1Group 1

AkronAkron LargeLarge

CantonCanton

MansfieldMansfield Group 1Group 1

SteubenvilleSteubenville

YoungstownYoungstown

Group 2 – Group 2 – removed because removed because of overlap with of overlap with LargeLarge

Dayton removed Dayton removed because of because of independent independent model model developmentdevelopment

Trip Length Analysis - Trip Length Analysis - PurposePurpose Intent of analysis was to find Intent of analysis was to find

areas where a friction factor areas where a friction factor curve could be shared between curve could be shared between areas.areas.

Same combination datasets were Same combination datasets were considered: small, large, group 1 considered: small, large, group 1 and group 2and group 2

Trip Length Analysis – Trip Length Analysis – Statistical AnalysisStatistical Analysis Trips used in analysis were restricted to those Trips used in analysis were restricted to those

with both trip ends within an MPO area and with both trip ends within an MPO area and with known locations of trip ends.with known locations of trip ends.

Rather than using reported trip length, the Rather than using reported trip length, the skimmed trip length was used in the analysis.skimmed trip length was used in the analysis.

ANOVA was used to compare average trip ANOVA was used to compare average trip length.length.

Trips were compared two waysTrips were compared two ways– Same trip purpose across areasSame trip purpose across areas– Purposes within an area to see if differences Purposes within an area to see if differences

existedexisted

Trip Length Analysis by Trip Length Analysis by Purpose - ResultsPurpose - Results Results indicate that there is significant Results indicate that there is significant

difference between average trip length difference between average trip length between areas in combination datasets.between areas in combination datasets.

Logical findings given different network Logical findings given different network characteristics, geographic size of area and characteristics, geographic size of area and other travel related factors.other travel related factors.

LargeLarge SmallSmall Group 1Group 1 Group 2Group 2

F-StatF-Stat SigSig F-StatF-Stat SigSig F-StatF-Stat SigSig F-StatF-Stat SigSig

HBWHBW 18.4718.47 0.000.00 124.81124.81 0.000.00 53.0953.09 0.000.00 36.7636.76 0.000.00

HBSHHBSH 24.1824.18 0.000.00 44.9244.92 0.000.00 10.6210.62 0.000.00 3.783.78 0.020.02

HBOHBO 60.3560.35 0.000.00 63.5263.52 0.000.00 8.218.21 0.000.00 73.8773.87 0.000.00

NHBWNHBW 12.5512.55 0.000.00 38.4238.42 0.000.00 10.5410.54 0.000.00 7.877.87 0.000.00

NHBONHBO 12.7712.77 0.000.00 34.8834.88 0.000.00 18.3718.37 0.000.00 11.5211.52 0.000.00

Average Trip Length by Average Trip Length by MPO AreaMPO Area

0

2

4

6

8

10

12

14

Ave

rag

e T

rip

Len

gth

HBW HBSH HBO NHBW NHBO

Trip Purpose

TOL

LIM

DAY

SPG

AKR

CAN

MAN

STE

YOU

Trip Length Frequency Trip Length Frequency Distribution - HBWDistribution - HBW

0%

10%

20%

30%P

ercen

tAkron Canton Dayton

Lima Mansfield Springfield

Steubenville Toledo Youngstown

0%

10%

20%

30%

Per

cen

t

20.00000 40.00000 60.00000 80.00000

Time

0%

10%

20%

30%

Per

cen

t

20.00000 40.00000 60.00000 80.00000

Time

20.00000 40.00000 60.00000 80.00000

Time

Trip Length Analysis by Trip Length Analysis by Area - ResultsArea - Results Results for comparison of purposes within an MPO Results for comparison of purposes within an MPO

area showed little potential for combination.area showed little potential for combination. Consistent with traditional approaches to have Consistent with traditional approaches to have

unique gravity model for each trip purpose.unique gravity model for each trip purpose.

ToledToledoo

HBWHBW HBSHHBSH HBOHBO NHBWNHBW NHBONHBO

FF SigSig FF SigSig FF SigSig FF SigSig FF SigSig

HBWHBW537.39537.39

00 0.0000.000741.27741.27

440.000.00

00118.08118.08

66 0.0000.000646.86646.86

99 0.0000.000

HBSHHBSH 30.48930.4890.000.00

00 87.88387.883 0.0000.000 4.7434.743 0.0290.029

HBOHBO 34.63334.633 0.0000.000 16.65716.657 0.0000.000

NHBWNHBW 65.63565.635 0.0000.000

NHBONHBO

Average Trip Length by Average Trip Length by Purpose by AreaPurpose by Area

0

2

4

6

8

10

12

14

Tol Lim Day Spr Akr Can Man Ste You

HBW

HBSH

HBO

NHBW

NHBO

Time of Day - PurposeTime of Day - Purpose

Determine whether Determine whether datasets could be datasets could be combined for estimation of combined for estimation of time of day factors and time of day factors and directional factors for Time directional factors for Time of Day model.of Day model.

Coincidence Ratio was Coincidence Ratio was used to determine if all used to determine if all areas shared similar daily areas shared similar daily distribution of trips.distribution of trips.

Four time periods were Four time periods were defined:defined:– Over Night (6pm to 6am)Over Night (6pm to 6am)– AM Peak Period (6am to AM Peak Period (6am to

9am)9am)– Midday (9am to 2pm)Midday (9am to 2pm)– PM Peak Period (2pm to PM Peak Period (2pm to

6pm)6pm)

REPORTED DEPARTURE TIME

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

14.0%

16.0%

18.0%

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

HOUR

PE

RC

EN

TA

GE

OF

TR

PS

HBW HBShop HBOther NHBWork NHBOther Combined

Overall Departure by Hour

0.00%

2.00%

4.00%

6.00%

8.00%

10.00%

12.00%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Departure Hour

Per

cen

tag

e o

f T

rip

s

MPO1 MPO2 MPO3 MPO4 MPO5 MPO6

MPO7 MPO8 MPO9 Combined

Time of Day – Time of Day – Statistical AnalysisStatistical Analysis Difference of proportions test was Difference of proportions test was

used to compare the proportion of used to compare the proportion of trips made between each area trips made between each area being compared:being compared:– Small with LargeSmall with Large– Small onlySmall only– Large onlyLarge only– Group 1 and Group 2Group 1 and Group 2

Time of Day - ResultsTime of Day - Results

From a cursory inspection, it seems all From a cursory inspection, it seems all areas could share the same dataset.areas could share the same dataset.

Further review of the results indicates that Further review of the results indicates that for HBSH (Period 1), HBW (Period 2), HBO for HBSH (Period 1), HBW (Period 2), HBO (Period 3) and HBSH and HBO (Period 4) (Period 3) and HBSH and HBO (Period 4) there are significant differences between there are significant differences between the small and large datasets.the small and large datasets.

Since all MPOs are included as either small Since all MPOs are included as either small or large, this was the recommended or large, this was the recommended dataset for TOD calibration.dataset for TOD calibration.

Time of Day - ResultsTime of Day - Results

0

10

20

30

40

1 2 3 4

Small Large

HBWHBW HBSHHBSH HBOHBO NHBWNHBW NHBONHBO

SmallSmall LargeLarge SmallSmall LargeLarge SmallSmall LargeLarge SmallSmall LargeLarge SmallSmall LargeLarge

Per 1Per 1 18.718.7 19.919.9 32.032.0 27.327.3 24.124.1 23.923.9 7.07.0 7.67.6 17.617.6 18.618.6

Per 2Per 2 35.935.9 32.832.8 3.73.7 4.24.2 23.223.2 23.323.3 17.917.9 17.417.4 9.69.6 8.68.6

Per 3Per 3 12.912.9 13.013.0 33.733.7 33.133.1 19.419.4 18.318.3 37.637.6 38.838.8 37.537.5 37.837.8

Per 4Per 4 32.532.5 34.334.3 30.630.6 35.435.4 33.333.3 34.534.5 37.537.5 36.136.1 35.335.3 35.135.1

HBW – Percent of Trips by Period

0

10

20

30

40

1 2 3 4

Small Large

HBSH – Percent of Trips by Period

Percent Departure by Period (Shaded = Statistically Different)

Additional AnalysisAdditional Analysis

Reviewed cell compression Reviewed cell compression scheme suggested by ODOT.scheme suggested by ODOT.– Cells compressed based on Cells compressed based on

rarity of households in rarity of households in surveysurvey

– Cells with more vehicles Cells with more vehicles than persons were than persons were compressed (based on compressed (based on analysis of OKI, MORPC and analysis of OKI, MORPC and NOACA survey)NOACA survey)

Evaluation of compression Evaluation of compression based on:based on:– Number of households in Number of households in

each cell from survey each cell from survey datasetdataset

– Difference in trip rate Difference in trip rate between independent cells between independent cells and compressed cellsand compressed cells

Analysis supported ODOT Analysis supported ODOT compression techniques.compression techniques.

0 0 WrkWrk

1 1 WrkWrk

2 2 WrkWrk

3 3 WrkWrk

0 0 VehVeh

1 1 VehVeh

2 2 VehVeh

3 3 VehVeh

1 HH1 HH 2 HH2 HH 3 HH3 HH 4 HH4 HH

0 0 VehVeh

1 1 VehVeh

2 2 VehVeh

3 3 VehVeh

Additional AnalysisAdditional Analysis

Evaluated the potential Evaluated the potential of a HB School trip of a HB School trip purpose.purpose.– Compared Average trip Compared Average trip

length for school and length for school and non school HB activitiesnon school HB activities

– Evaluated frequency of Evaluated frequency of trips for sufficient trips for sufficient numbers for calibration.numbers for calibration.

– Evaluated distribution Evaluated distribution of households in cross of households in cross classification matrix classification matrix (vehicle ownership x (vehicle ownership x students in household)students in household)

Determined that a HB Determined that a HB School purpose was School purpose was warrantedwarranted

Large Combined Dataset

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75

Trip Length

Per

cen

t o

f T

rip

s

L_HBS

L_HBO`

MPOMPO Survey HHSurvey HH HH w/Sch TripHH w/Sch Trip

TOLTOL 21762176 597597

LIMLIM 13281328 302302

DAYDAY 19501950 521521

SPRSPR 13491349 394394

AKRAKR 19361936 559559

CANCAN 13191319 351351

MANMAN 13041304 332332

MANMAN 12761276 249249

YOUYOU 12511251 324324

Trip Rate Sensitivity AnalysisTrip Rate Sensitivity Analysis

Further pursue the impact that Further pursue the impact that various trip generation rates would various trip generation rates would have on model resultshave on model results

Calculate various “feasible” sets of Calculate various “feasible” sets of trip rates based on the combined trip rates based on the combined and Toledo stand-alone survey data and Toledo stand-alone survey data setssets

Smaller sample size in the stand Smaller sample size in the stand alone data implies a broader range alone data implies a broader range of “feasible” trip rate setsof “feasible” trip rate sets

Total Households ComparisonTotal Households Comparison

Combined Akron-ToledoWRKER

VEH 0 1 2 30 129 281 648 656 692 346 592 826 243 69 210 356 159 4112

ToledoWRKER

VEH 0 1 2 30 65 111 377 373 462 173 294 444 153 22 102 167 87 2176

Trip RatesTrip RatesCombined Areas-Base Rates

HBW V0W0 V0W1 V0W2 V0W3 V1W0 V1W1 V1W2 V1W3 V2W0 V2W1 V2W2 V2W3 V3W0 V3W1 V3W2 V3W3

CU 0.00 0.95 0.95 0.95 0.02 1.09 3.22 3.22 0.03 1.25 2.34 3.13 0.05 1.14 2.54 3.66

SR 0.00 0.96 0.96 0.96 0.02 1.11 2.49 2.49 0.05 1.10 2.46 1.75 0.05 1.26 2.43 4.01

NHBW V0W0 V0W1 V0W2 V0W3 V1W0 V1W1 V1W2 V1W3 V2W0 V2W1 V2W2 V2W3 V3W0 V3W1 V3W2 V3W3

CU 0.00 0.32 0.32 0.32 0.01 0.60 0.57 0.60 0.02 0.51 1.13 0.99 0.00 0.48 1.01 1.38

SR 0.00 0.35 0.35 0.35 0.01 0.53 1.05 1.05 0.04 0.59 1.32 0.59 0.06 0.43 1.13 1.42

HBO V0H1 V0H2 V0H3 V0H4 V1H1 V1H2 V1H3 V1H4 V2H1 V2H2 V2H3 V2H4 V3H1 V3H2 V3H3 V3H4

CU 0.92 2.39 2.39 2.39 1.11 2.37 3.57 3.57 1.11 2.44 3.65 6.38 1.11 2.44 3.09 6.49

SR 0.39 1.51 1.51 1.51 1.16 1.83 3.76 3.76 1.16 2.51 3.22 6.11 1.16 2.51 3.21 5.44

HBSH V0H1 V0H2 V0H3 V0H4 V1H1 V1H2 V1H3 V1H4 V2H1 V2H2 V2H3 V2H4 V3H1 V3H2 V3H3 V3H4

CU 0.23 0.57 0.57 0.57 0.43 0.66 1.56 1.56 0.43 0.69 0.85 1.21 0.43 0.69 1.01 1.07

SR 0.28 1.23 1.23 1.23 0.40 0.67 0.71 0.71 0.40 0.87 0.75 1.03 0.40 0.87 0.81 1.24

NHBO V0H1 V0H2 V0H3 V0H4 V1H1 V1H2 V1H3 V1H4 V2H1 V2H2 V2H3 V2H4 V3H1 V3H2 V3H3 V3H4

CU 0.63 1.50 1.50 1.50 0.85 1.10 1.92 1.92 0.85 1.27 1.68 3.12 0.85 1.27 2.13 2.98

SR 0.09 0.98 0.98 0.98 0.93 1.13 2.14 2.14 0.93 1.52 1.54 2.91 0.93 1.52 1.99 2.64

HBSC V0C0 V0C1 V0C2 V0C3 V1C0 V1C1 V1C2 V1C3 V2C0 V2C1 V2C2 V2C3 V3C0 V3C1 V3C2 V3C3

CU 0.00 1.80 1.80 1.80 0.00 1.23 3.22 3.22 0.00 1.15 2.79 4.65 0.00 1.15 2.62 4.69

SR 0.00 2.40 2.40 2.40 0.00 0.91 3.11 3.11 0.00 1.19 2.80 4.20 0.00 1.19 3.12 4.70

CBD/Urban HBW Trip Rates

0

0.5

1

1.5

2

2.5

3

3.5

4

V0W0

V0W1

V0W2

V0W3

V1W0

V1W1

V1W2

V1W3

V2W0

V2W1

V2W2

V2W3

V3W0

V3W1

V3W2

V3W3

Combined

ToledoOnly

Suburban/Rural HBW Trip Rates

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

V0W0

V0W1

V0W2

V0W3

V1W0

V1W1

V1W2

V1W3

V2W0

V2W1

V2W2

V2W3

V3W0

V3W1

V3W2

V3W3

Combined

Toledo Only

Construction of Alternate Construction of Alternate Trip RatesTrip Rates

Calculate Percent Errors for a Calculate Percent Errors for a given confidence interval (rather given confidence interval (rather arbitrarily selected 90%)arbitrarily selected 90%)

E = Z*CV/SQRT(N)E = Z*CV/SQRT(N) Develop other feasible sets of trip Develop other feasible sets of trip

rates within plus / minus this error rates within plus / minus this error percentage of the calculated percentage of the calculated meanmean

CBD/Urban HBW %Error at 90% Confidence

0%

20%

40%

60%

80%

100%

120%

140%V

0W0

V0W

1V

0W2

V0W

3V

1W0

V1W

1V

1W2

V1W

3V

2W0

V2W

1V

2W2

V2W

3V

3W0

V3W

1V

3W2

V3W

3

Combined

Toledo Only

Suburban/Rural HBW %Error at 90% Confidence

0%

20%

40%

60%

80%

100%

120%

140%

160%V

0W0

V0W

1V

0W2

V0W

3V

1W0

V1W

1V

1W2

V1W

3V

2W0

V2W

1V

2W2

V2W

3V

3W0

V3W

1V

3W2

V3W

3

Combined

Toledo Only

Construction of Alternate Construction of Alternate Trip RatesTrip Rates

Trip rates varied by cross-class cell, however, Trip rates varied by cross-class cell, however, the overall resultant trip rates were also held the overall resultant trip rates were also held within the 90% confidence intervalwithin the 90% confidence interval

Various perturbations of the trip rates were Various perturbations of the trip rates were created within this range, the two shown are:created within this range, the two shown are:– Systematic perturbation involving increasing zero Systematic perturbation involving increasing zero

Vehicle HH trip rates by exactly the calculated Vehicle HH trip rates by exactly the calculated percent error while reducing all other trip rates by percent error while reducing all other trip rates by 10% of this value10% of this value

– Random perturbation of each trip rate within its Random perturbation of each trip rate within its percent error rangepercent error range

HBW Proportion of the Percent Error HBW Proportion of the Percent Error Applied to Create Alternate Trip RateApplied to Create Alternate Trip Rate

Fraction to Apply (Case 1,Systematic Error) Fraction to Apply (Case 2, Random Error)CU SR CU SR

V0W0 -1 -1 V0W0 0.83 -0.13V0W1 -1 -1 V0W1 -0.43 -0.56V0W2 -1 -1 V0W2 0.01 0.38V0W3 -1 -1 V0W3 0.61 0.82V1W0 0.1 0.1 V1W0 -0.36 -0.06V1W1 0.1 0.1 V1W1 -0.96 -0.22V1W2 0.1 0.1 V1W2 0.69 0.37V1W3 0.1 0.1 V1W3 -0.43 -0.29V2W0 0.1 0.1 V2W0 0.41 0.82V2W1 0.1 0.1 V2W1 -0.69 -0.46V2W2 0.1 0.1 V2W2 0.97 -0.31V2W3 0.1 0.1 V2W3 0.49 -0.22V3W0 0.1 0.1 V3W0 0.89 -0.98V3W1 0.1 0.1 V3W1 -0.07 -0.17V3W2 0.1 0.1 V3W2 0.38 -0.54V3W3 0.1 0.1 V3W3 0.90 0.54

Alternate Reality Socio-Alternate Reality Socio-Economic DataEconomic Data

Given the concentration of variance in Given the concentration of variance in certain rare cells of the cross classificatin certain rare cells of the cross classificatin matrix…matrix…

An alternative set of zonal SE data was An alternative set of zonal SE data was constructed that placed more HH’s in constructed that placed more HH’s in these cells by:these cells by:– Reducing Vehicles by 50% in CBD / Urban AreaReducing Vehicles by 50% in CBD / Urban Area– Increase Workers 16% in all zonesIncrease Workers 16% in all zones– No change in # of HH’s or attraction variablesNo change in # of HH’s or attraction variables

CBD/Urban HH Distributions

0

5000

10000

15000

20000

25000

V0W0

V0W1

V0W2

V0W3

V1W0

V1W1

V1W2

V1W3

V2W0

V2W1

V2W2

V2W3

V3W0

V3W1

V3W2

V3W3

Base Inputs

Mod Inputs

Suburban/Rural HH Distributions

0

5000

10000

15000

20000

25000

30000

V0W0

V0W1

V0W2

V0W3

V1W0

V1W1

V1W2

V1W3

V2W0

V2W1

V2W2

V2W3

V3W0

V3W1

V3W2

V3W3

Base Inputs

Mod Inputs

Test Impact on Test Impact on Measures of Measures of Effectiveness (MOE’s)Effectiveness (MOE’s) 12 Test Cases Based Upon:12 Test Cases Based Upon:

6 Sets of Trip Rates6 Sets of Trip Rates1.1. Combined Data, BaseCombined Data, Base2.2. Combined Data, Systematic PerturbationCombined Data, Systematic Perturbation3.3. Combined Data, Random PerturbationCombined Data, Random Perturbation4.4. Toledo Data, BaseToledo Data, Base5.5. Toledo Data, Systematic PerturbationToledo Data, Systematic Perturbation6.6. Toledo Data, Random PerturbationToledo Data, Random Perturbation

2 Sets of SE Data2 Sets of SE Data1.1. BaseBase2.2. ModifiedModified

Test Impact on Test Impact on Measures of Measures of Effectiveness (MOE’s)Effectiveness (MOE’s) Evaluate Various MOE’s:Evaluate Various MOE’s:

1.1. Link VolumeLink Volume

2.2. VMTVMT

3.3. VHTVHT

4.4. %RMSE or %RMSD%RMSE or %RMSD

5.5. Tons of PollutantsTons of Pollutants

6.6. TripsTrips

7.7. Transit RidersTransit Riders

Base Model & SE Data

Toledo Data Only,

Systematic Perturbation,

Modified SE Data

Volume on New River Crossing

VMTVMT

SE Data Survey Dataset Perturbation VMT FWY VMT ART VMT TOTBase Combined Area None 6499938 9805614 16305535

Systematic 6489457 9821983 16311461Random 6563521 9861934 16425457

Toledo Stand Alone None 6524711 9788532 16313243Systematic 6522553 9768231 16290776Random 6514602 9865388 16380002

Modified Combined Area None 6694326 10418412 17112774Systematic 6639266 10341460 16980756Random 6783347 10494852 17278214

Toledo Stand Alone None 6728134 10407264 17135398Systematic 6635597 10215620 16851224Random 6735962 10485054 17221020

%Difference in VMT With Respect to Base Trip Rates Using Base SE Data

-0.20%

-0.10%

0.00%

0.10%

0.20%

0.30%

0.40%

0.50%

0.60%

0.70%

0.80%

Com-Pert1 Com-Pert2 Tol-Base Tol-Pert1 Tol-Pert2

%Difference in VMT Between Demographic Scenarios

0%

1%

2%

3%

4%

5%

6%

Com-Base

Com-Pert1

Com-Pert2

Tol-Base Tol-Pert1 Tol-Pert2

VHTVHT

SE Data Survey Dataset Perturbation VHTBase Combined Area None 418502

Systematic 418902Random 422627

Toledo Stand Alone None 418675Systematic 417417Random 421897

Modified Combined Area None 445675Systematic 440379Random 451113

Toledo Stand Alone None 446977Systematic 434776Random 449970

%Difference in VHT With Respect to Base Trip Rates Using Base SE Data

-0.40%

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

1.20%

Com-Pert1 Com-Pert2 Tol-Base Tol-Pert1 Tol-Pert2

%Difference in VHT Between Demographic Scenarios

0%

1%

2%

3%

4%

5%

6%

7%

8%

Com-Base

Com-Pert1

Com-Pert2

Tol-Base Tol-Pert1 Tol-Pert2

%RMS Error and %RMS Error and DifferenceDifference

SE Data Survey Dataset Perturbation %RMSE %RMSDBase Combined Area None 43.19

Systematic 43.02Random 43.27

Toledo Stand Alone None 43.04Systematic 42.96Random 43.36

Modified Combined Area None 10.43Systematic 8.36Random 10.81

Toledo Stand Alone None 10.65Systematic 7.09Random 11.05

%Difference in %RMSE With Respect to Base Trip Rates Using Base SE Data

-0.60%

-0.40%

-0.20%

0.00%

0.20%

0.40%

0.60%

Com-Pert1 Com-Pert2 Tol-Base Tol-Pert1 Tol-Pert2%RMSD Between Demographic Scenarios

0%

2%

4%

6%

8%

10%

12%

Com-Base

Com-Pert1

Com-Pert2

Tol-Base Tol-Pert1 Tol-Pert2

Ozone PrecursorsOzone Precursors

SE Data Survey Dataset Perturbation HC (TONS)NOX (TONS)Base Combined Area None 32.67 55.71

Systematic 32.68 55.72Random 32.93 56.15

Toledo Stand Alone None 32.68 55.76Systematic 32.62 55.69Random 32.85 55.96

Modified Combined Area None 34.44 58.35Systematic 34.13 57.89Random 34.80 58.95

Toledo Stand Alone None 34.50 58.46Systematic 33.81 57.49Random 34.69 58.73

%Difference in Pollutants With Respect to Base Trip Rates Using Base SE Data

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

Com-Pert1

Com-Pert2

Tol-Base Tol-Pert1 Tol-Pert2

HC (TONS)

NOX (TONS) %Difference in Pollutants Between Demographic Scenarios

0%

1%

2%

3%

4%

5%

6%

Com-Base

Com-Pert1

Com-Pert2

Tol-Base

Tol-Pert1

Tol-Pert2

HC (TONS)

NOX (TONS)

TripsTrips

SE Data Survey Dataset Perturbation Total Trips Transit TripsBase Combined Area None 1,961,171 13442

Systematic 1,954,990 13314Random 1,969,270 13803

Toledo Stand Alone None 1,949,646 13350Systematic 1,938,103 13113Random 1,957,234 13734

Modified Combined Area None 2,117,025 14241Systematic 2,063,020 13623Random 2,125,166 14850

Toledo Stand Alone None 2,106,978 14199Systematic 2,015,572 13120Random 2,108,702 14776

%Difference in Trips With Respect to Base Trip Rates Using Base SE Data

-3.00%

-2.00%

-1.00%

0.00%

1.00%

2.00%

3.00%

Com-Pert1

Com-Pert2

Tol-Base Tol-Pert1 Tol-Pert2

Total Trips

Transit Trips %Difference in Trips Between Demographic Scenarios

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

Com-Base

Com-Pert1

Com-Pert2

Tol-Base Tol-Pert1 Tol-Pert2

Total Trips

Transit Trips

ConclusionsConclusions

Randomly perturbed trip rates, even when applied Randomly perturbed trip rates, even when applied to purposefully skewed SE data showed almost no to purposefully skewed SE data showed almost no impact on typical MOE’simpact on typical MOE’s

Systematically perturbed trip rates produced Systematically perturbed trip rates produced slightly lower %RMSD between the SE data slightly lower %RMSD between the SE data scenariosscenarios– Base %RMSD: Base %RMSD: 10.4310.43– Combined:Combined: 8.368.36– Stand Alone:Stand Alone: 7.09 7.09

These slight differences are minor compared to These slight differences are minor compared to the models %RMSE valuesthe models %RMSE values

ConclusionsConclusions

The Toledo stand alone sample was sufficient for The Toledo stand alone sample was sufficient for the given model (not surprising since it was the given model (not surprising since it was designed as such)designed as such)

Increasing sample size much beyond the computed Increasing sample size much beyond the computed minimums wouldn’t have added muchminimums wouldn’t have added much

It was still useful to combine the data sets where It was still useful to combine the data sets where practical to give more faith in the low incidence practical to give more faith in the low incidence cellscells

This also allowed the addition of the area type This also allowed the addition of the area type dimension to the smaller areas whose smaller dimension to the smaller areas whose smaller survey sample was not originally designed for thissurvey sample was not originally designed for this

Questions?Questions?

Please use the Microphone.Please use the Microphone.