Transcript Slide 1

In this presentation, we will:
1.
•
•
Describe each step the Compass model and show
comparable steps in the IRM.
Compass = What, , Where, How
IRM= Who, What, When, Where, How, Why
1.
Take a household through these steps and show how the
household is treated differently in the two models.
2.
Show how the additional complexity of the IRM provides
additional sensitivity (we can test and represent more.)
Reality Is more complicated than 4 step models can show.
–
“Putting on the Glasses”
The New Regional Model Focus
Big Picture
• Effects of development pattern/design
– UGB scenarios
– TOD
– Urban Centers
– “New” Urbanism vs traditional 20th century
suburban patterns
– Mixed versus single use
The New Regional Model Focus
Big Picture
• Explicit modeling of bike/ped.
• Time of day modeling
• Person/household characteristics:
– Age
– Gender
– Household population composition
• Presence and age of children
• Number of workers
The New Regional Model Focus
Big Picture
• Number of vehicles
• Work at home
• Full induced demand for the first time
– Trip/tour suppression/re-structuring
– Trip-length changes
– Mode changes
– Time of day changes
– Path changes
The New Regional Model Focus
Details for Tekkies
• Modeling trips in tours, not separately
• Each household and job is given a precise
location (xy point)
• Each household/person in the region is
represented individually – PUMS-level
data
The New Regional Model Focus
Details for Tekkies
• More / real trip purposes:
– Old model: home-based work, home-based
non-work, non-home-based
– New model: work, school, escort, shopping,
eat meal, social-recreation, personal business
The New Regional Model Focus
Bottom Line
• The model operates at the level at which
decision actually are made
– Usually the person
– Occasionally the household
The New Regional Model Focus
Bottom Line
• The model operates at the level at which
decision actually are made
– Usually the person
– Occasionally the household
The New Regional Model Focus
Bottom Line
• disaggregate modeling means recognizing
that things are different, and describing
them as they are:
– Person/household type
– Tour/trip type
– Location
– Time of day
The New Regional Model Focus
Bottom Line
• better modeling now – better ability to
improve in the future – better ability to run
scenarios
– Example – where will the elderly live?
– Example – use of hybrid / electric vehicles.
Compass
i. Network Processing
ii. Area Type
1. Trip Generation
i. Highway/Transit Skims
2. Trip Distribution
3. Mode Choice
i. Parking Cost
ii. Time-of-Day
4. Highway/Transit Assignment
IRM
1. Population Synthesizer
Network Skims
Aggregate Mode/Destination Choice Logsum Generator
Mode Choice Logsum Generator
2. Regular Workplace Location
Choice
Exact Number of Tours Choice
Work Tour Destination Type Choice Model
Work-Based Subtour Generation Choice
3. Regular School Location
Choice
6. Tour Primary Destination
Choice
4. Auto Availability
7. Tour Main Mode Choice
Intermediate Stop Logsum Generator
5. Daily Activity Pattern Choice
8. Tour Time of Day Choice
Intermediate Stop Generation Choice
Intermediate Stop Location Choice
9. Trip Mode Choice
10. Trip Departure Time Choice
11. Assignment
How does the model do its job?
It answers these questions:
–
–
–
–
–
–
–
–
–
–
What kind of people…
In what kinds of households…
Living in what locations...
Working and going to school where and how
Do what kinds of things during their day…
Requiring them to make what kinds of trips…
To what locations…
By what travel modes…
By what paths through the transportation system?
At what times?
How does the model do its job?
It answers these questions:
–What kind of people…
–
–
–
–
–
–
–
–
–
In what kinds of households…
Living in what locations…
Working and going to school where and how
Do what kinds of things during their day…
Requiring them to make what kinds of trips…
To what locations…
By what travel modes…
By what paths through the transportation system?
At what times?
What do we know about the people in the
households in the Compass model?
Answer: nothing.
? ??
Example Family:
Adult - Age?
Job status?
Adult – Age?
Job status?
Child – Age?
Student status?
Relationship
To adults?
What do we know about the people in the
households in the IRM?
Answer: Anything included in the Census.
Example Family:
Mother, Age 33
Part Time Service
Worker
Father, Age 34
Full Time Education
Worker
Son, Age 4
Pre-School Student
Family Income :
$61,000
What do we know about
individual people in Compass?
Person Data in the IRM
Household
Person ID ID
Age
556053
321005
556054
321006
556055
321007
Person ID
556053
556054
556055
School
Grade
29
8
61
Relationship to
Race
Householder
Afr American Householder
White
Son
Asian
Householder
Gender
0 Female
2 Male
0 Male
Weeks
School Worked per
Grade
Year
0
52
2
0
0
32
Hours
Worked Per Income
Week
Earned
40
23400
1
0
16
32500
Marital
Status
Married
Single
Single
Poverty
Status
Employ
Category
Service
0
Retail
Employ
Status
0 Full Time
1 Not
0 Part Time
How does the model do its job?
It answers these questions:
– What kind of people…
–In what kinds of households…
–
–
–
–
–
–
–
Living in what locations...
Working and going to school where and how
Do what kinds of things during their day…
Requiring them to make what kinds of trips…
To what locations…
By what travel modes…
By what paths through the transportation system?
Household Data in Compass
Household Data in the IRM
Number in
Household Number in
Household Number of Number of Under Age Household
Income
People
Children
5
age 5 to 15
Household
ID
TAZ
342652
342653
342654
307020
307020
307020
Years at
Residence Race
6
1
3
55000
51000
56000
0
1
4
1
2
6
0
0
0
0
1
2
0
0
2
Distance
from
household Number of
to nearest Full Time
transit stop Workers
Number of
University Own or
Students
Rent
Building
Size
1
3
2
1
2
6
1
0
1
0.5
1
2
1
1
2
How does the model do its job?
It answers these questions:
– What kind of people…
– In what kinds of households…
–Living in what locations...
–
–
–
–
–
–
–
Working and going to school where and how
Do what kinds of things during their day…
Requiring them to make what kinds of trips…
To what locations…
By what travel modes…
By what paths through the transportation system?
At what times?
704160
705110
301470
302250
301460
705140
Ok, so what do we know about
household locations
704190 ?
in Compass
090
704180
705170
301500
705180
301480
306010
Answer:704200
They705150
are705160
located in traffic
301520 analysis zones,
704110
and…
301510 301490 306020
704210
301550
704220
120th Ave
704230
307070 307080
307020
704130
307010
307040 307050
202120
307060
307100 307090
307030
202130
36
US
2260
307150
307140
307120
307130
202270
202290
202300
307160
307170
307180
307270
202280
307110
I-25
202140
306210
307240
307220
307190
306230
And what do we know about
households in the IRM?
We know a lot more about
households in the IRM: why is that
better?
• Households with more drivers and workers own
more cars.
• Households with more cars make different
choices than households with fewer cars:
– They make more tours
– And use drive mode for them more often
• Point-level location means we actually know:
–
–
–
–
Walk distance to/from transit
Walk trip distance
Bike trip distance
Short auto trip distance.
We know a lot more about people
in the IRM:
why is that better?
• Lots of reasons!
– People tend to work in places where there are
a lot of jobs in their field.
– Kids tend to go to school where their older
siblings go.
– Workers tend to go to work, students tend to
go to school, retired people tend not to do
either (etc.)
– People with kids tend to cart them around a
lot, and drive doing it.
How does the model do its job?
It answers these questions:
– What kind of people…
– In what kinds of households…
– Living in what locations...
–Working and going to school
where and how
–
–
–
–
–
–
Do what kinds of things during their day…
Requiring them to make what kinds of trips…
To what locations…
By what travel modes…
By what paths through the transportation system?
At what times
412223
412224
412321
412312
What do we know
about where jobs are
in Compass?
Answer: they are in
traffic analysis
zones, and…
412204 412205 412206
412211
41219
412212
412214
412322
412324
412215
412213 412207 412208 412209
412323
Colfax Avenue
Civic Center Park
412340
41233
412380
4123
412360
Broadway
412350
405080
405090
412390
What do we
know about
where jobs are
in the IRM?
Job Data in Compass
Job Data in the IRM
Building ID
115497
18663
115498
Legal Name
Colorado State Glass and Mirror
Lookin Good Day Spa
Bruner Consulting Inc
Employment
Type
Retail
Service
Service
Building ID X Coordinate Y Coordinate
Wages
115497
5689090
4572891
$62,300
18663
4089972
57890849
$5,200
115498
5575757
2890798
$6,700
Number of
Employees
Address
9 1120 Cooke Ct
2 6662 W 38th Ave
1 1117 Alder Way
What do we know about
schools in Compass?
And about Schools in the IRM?
We know a lot more in the IRM:
why is that better?
• Tie the kind of job people have to the kind
of company and its location.
• We know precisely how far the company is
from the transit stop.
• Non-university students tend to go school
in their home school district.
• We know precisely how far the school is
from each home.
How does the model do its job?
It answers these questions:
–
–
–
–
What kind of people…
In what kinds of households…
Living in what locations...
Working and going to school where and how
–Do what kinds of things during
their day…
–Requiring them to make what
kinds of trips…
–
–
–
–
To what locations…
By what travel modes…
By what paths through the transportation system?
At what times?
What do we know about what people
do during their day in Compass?
• Number of work trips by households in the
zone.
• Number of other kinds of trips made by
households in the zone.
Compass Trip Rates
Note also: only three types of trips- Home-Based Work, Home-Based
Non-Work, and Non-Home Based
Table 13: Final HBW Trip Production Rates
Household Size
1
2
3
4
Low (Less than $15,000)
0.515
1.339
1.339
1.339
Middle ($15,000 - $74,999)
1.236
1.854
2.060
2.266
High ($75,000 or more)
0.927
1.854
2.472
2.472
Source: PRD_HB_R.asc TransCAD file dated 10/22/2003
Income Group
5+
2.575
2.575
2.060
Table 14: Final HNW Trip Production Rates
Household Size
1
2
3
4
Low (Less than $15,000)
1.272
2.438
4.982
4.982
Middle ($15,000 - $74,999)
1.378
3.392
4.876
7.526
High ($75,000 or more)
1.590
2.968
4.876
9.540
Source: PRD_HB_R.asc TransCAD file dated 10/22/2003
Income Group
5+
4.982
10.282
10.282
Table 15: Final NHB Trip Production Rates
Household Size
1
2
3
4
Low (Less than $15,000)
0.784
0.784
2.912
2.912
Middle ($15,000 - $74,999)
1.344
2.576
2.912
3.808
High ($75,000 or more)
2.240
3.248
4.256
4.816
Source: PRD_HB_R.asc TransCAD file dated 10/22/2003
Income Group
5+
2.912
3.584
5.936
What do we know in the IRM?
• Mostly about things they do when they
leave the house:
– Exception: we know if they work at home.
• Out-of-home activities: work, school, shop,
eat meal, socialize, escort others, personal
business.
IRM Activities in the Day
Doing what?
Working
Primary reason
for going out
Just
making a
stop along
the way
x
Going to school
Shopping
Escorting others
x
Socializing
Eating out
Personal business
x
Tours Generated in the IRM
PARK
AND
RIDE
HOME
DAY
CARE
TOUR-BASED MODEL
1 home-based work tour
•1 work- based meal tour
•2 intermediate serve passenger
stops
TRIP-BASED MODEL
•4 non-home based trips
Walk
•Two home-based other trips
•NHB trip poorly handled...
WORK
RESTAURANT
IRM Activities in the Day
Doing what?
Working
Primary reason
for going out
Just
making a
stop along
the way
x
Going to school
X
Shopping
Escorting others
Socializing
Eating out
Personal business
x
Tours Generated in the IRM
HOME
WORK
STORE
TOUR-BASED MODEL
TRIP-BASED MODEL
•1 home-based work tour
• 1 home-based work trip
•1 shopping stop
•1 non-home-based trip
•1 home-based non work trip
IRM Activities in the Day
Doing what?
Primary reason
for going out
Working
Going to school
Shopping
Escorting others
Socializing
Eating out
Personal business
X
Just making
a stop along
the way
Tours Generated in the IRM
passenger
HOME
DAY CARE
TOUR-BASED MODEL
TRIP-BASED MODEL
•1 school tour
• 2 home-based non work trips
TOTAL TOURS BY INDIVIDUAL:
13 TOTAL TRIPS BY HOUSEHOLD:
WOMAN: 1 HOME BASED WORK TOUR
1 HOME-BASED WORK
MAN:
1 WORK-BASED MEAL TOUR
5 HOME-BASED NON-WORK
2 SERVE PASSENGER STOPS
7 NON-HOME BASED:
1 HOME BASED WORK TOUR
1 SHOPPING STOP
CHILD: 1 HOME-BASED SCHOOL TOUR
Why is it better that the IRM is
more detailed in describing why
people travel?
• Able to depict how changes in
demographics, like a larger older
population, can cause different amounts
and types of travel.
• Able to represent how much accessibility
and mixed use density a person’s home
zone has to other locations impact the
amount of travel they do.
How does the model do its job?
It answers these questions:
–
–
–
–
–
–
What kind of people…
In what kinds of households…
Living in what locations...
Working and going to school where and how
Do what kinds of things during their day…
Requiring them to make what kinds of trips…
–To what locations…
– By what travel modes…
– By what paths through the transportation system?
– At what times?
To what location do people go to do the
activities? In Compass, the trip attractions
and productions are used to predict trip origins
and destinations.
Table 23: Final HBW Trip Attraction Rates by Income Group
Trip Attraction Rate per:
Household
Production /
Distribution
Employee
Retail
Employee
Service
Employee
Low (Less than $15,000)
0.003
0.027
0.082
0.070
Middle ($15,000 - $74,999)
0.020
0.787
0.916
0.776
High ($75,000 or more)
0.014
0.269
0.250
0.440
Total
0.037
1.083
1.248
1.286
Income Group
Source: ATTHBW_R.asc TransCAD file dated 10/22/2003
1 Total rates shown only for comparison with models from other regions.
Notes:
In Trip-based models, the gravity model predicts the number of trips from
origin to destination based on the number of productions in the origin zone
and attractions in the destination zone.
Friction Factors are calibrated so that modeled trip length frequency
distributions match observed trip length frequency distributions.


 K F A

ij
ij
j

Tij  Pi   n
 K  F  A  
ik
ik
k 
 
k 1

Tij
Pi
Aj
Kij
Fij
i
j
n
=
=
=
trips between TAZ i and TAZ j
=
productions in TAZ i
=
attractions in TAZ j
“K-factor” adjustment between TAZ i and TAZ j
“friction factor” between TAZ i and TAZ j
=
production TAZ
=
attraction TAZ
=
total number of TAZs
Trip Distribution Output: O-D matrix
Compass model output
From Trip Distribution:
Home-Based Work Trips from Zone 307020
Compass model output
From Trip Distribution:
Home-Based Non-Work Trips from Zone 307020
Where does the woman go during her day? The IRM
destination choice models could predict the following for
the woman:
Tour origin
Tour
destination
The IRM destination choice models could
predict the following for the man:
Why is it better how destinations
are chosen in the IRM?
• Can test how desirable a location is by
how easy it is to get there by all modes
including transit.
• Can test how mixed use density causes a
destination to be more desirable
• Can test how a person chooses
destinations close to their usual work or
school zone.
How does the model do its job?
It answers these questions:
–
–
–
–
–
–
–
What kind of people…
In what kinds of households…
Living in what locations...
Working and going to school where and how
Do what kinds of things during their day…
Requiring them to make what kinds of trips…
To what locations…
–By what travel modes…
– By what paths through the transportation system?
– At what times?
How does mode choice work in
Compass?
• Each trip purpose has a model (HBW,
HBNW, NHB.)
• All trips between each zone-pair are
treated as being identical.
• Trips that are really in the same tour know
nothing about each other.
• Outputs are trip tables by mode by
purpose (and by income for HBW.)
Compass mode choice output:
O-D Trips by Mode
PARK
AND
RIDE
DriveSR2
HOME
DAY
CARE
Example IRM
Mode Choices
for the Woman
Walk
RESTAURANT
WORK
HOME
Example IRM
Mode Choices
for the Man
Drive Alone
STORE
WORK
passenger
DAY CARE
HOME
Drive- SR2
Example IRM
Mode Choices
for the Child
Why is the IRM better at
representing mode choice?
• Has bike and walk modes as a choice
• Represents how Origin and destination
employment density impacts a person’s
mode choice
• Represents how a person’s auto availability
impacts mode choice
• Allows school tours and work-based
subtours to have different mode choices
than other tours (Compass just has homebased other)
How does the model do its job?
It answers these questions:
–
–
–
–
–
–
–
–
–
What kind of people…
In what kinds of households…
Living in what locations...
Working and going to school where and how
Do what kinds of things during their day…
Requiring them to make what kinds of trips…
To what locations…
By what travel modes…
By what paths through the transportation system?
–At what times?
When do trips occur?
Compass
Very Simple Model:
Time-of-Day Factors
For Example,
68% of Home-Based Work Tours, arrive at work from
6:30 am –9:00 am.
Compass: Time-of-Day Factors applied
(based on total observed vehicles hours observed in each period)
AM1
AM2
AM3
800
OP1
700
100,000
OP3
OP2
PM1
PM2
PM3
OP4
OP1
90,000
70,000
60,000
50,000
40,000
30,000
20,000
10,000
Time Period
2300
2200
2100
2000
1900
1800
1700
1600
1500
1400
1300
1200
1100
1000
900
600
500
400
300
200
100
0
0
Expaned Vehicle Hours Traveled
80,000
Compass Model Time of Day
Outputs: O-D Trips by Time of Day
IRM Time of Day Models
• Tour time-of-day:
– Predicts start and end of tour
– Higher priority tours run first, block out times
of day not available to lower priority toursTour
time-of-day:
• Trip time-of-day:
– Predicts departure time from each stop
– In-transit time known, so serves as departure
time and duration model
PARK
AND
RIDE
time
HOME
DAY
CARE
Example IRM
Mode Choices
for the Woman
time
RESTAURANT
WORK
Why are the IRM time of day
models better?
• Can better represent time-shifting due to
congestion
• Can represent how changes in
demographics impact time of day choices,
i.e. more retired people means less
congestion peaking
• Can represent how a person’s mode
choice impacts what time of day they
travel, i.e. I can’t take the bus until 3:15 PM
Example Tour Outcomes for the
Family : Purpose, Origin, Destination
Mode, Time of Day
Tour ID
541298
541299
541300
541898
Person ID
127804
127804
127805
127806
Tour Origin
Tour Purpose Point ID
Work
Restaurant
School
Work
987130
1384689
987130
987130
Tour
Destination
Point ID
1384689
1578908
693920
578925
Tour Mode
Drive to Transit
Walk
Shared Ride 2
Drive Alone
Tour
Destination
Arrival Time
8:00 AM
12:00 PM
7:00 AM
8:00 AM
Tour
Destination
Departure
Time
5:00 PM
1:00 PM
6:00 PM
5:00 PM
Finally, in both the IRM and Compass models the
choices of where to go, when, by what mode are
assigned to the networks.
Example AM Peak Highway Flows:
The woman drives with the child to the daycare down
120th, then takes the bus route 122X to Civic Center, and
walks from the station to work down Broadway.
Key Model Differences
•
•
•
•
Trip/tour generation sensitivity.
Time-of-day sensitivity.
Development pattern sensitivity.
Modeling of non-motorized modes and
walk access to transit.
• Trips connected in tours.
• Person-level decisions made at the person
level – using many person characteristics.
Scenario Sensitivity ExamplesWhy the IRM is better?
How does putting on the glasses help us see better?
• Effects of aging population or other changes in
demographics.
• Effects of neighborhood “gentrification.”
• Enhanced EJ evaluation.
• Better evaluation of “induced demand.”
• “Peak spreading” effects.
• Effects of transit-oriented development.
• Effect of built environment on Bike and Ped
Movement
Walk Mode Shares
Example Output
Bike Mode Shares
Example Output
Factors that make a trip
more likely to use
walk/bike modes
Factors that make a trip
less likely to use
walk/bike modes
PERSON CHARACTERISTICS
•No car in Household
•Fewer cars than drivers
•Low income
•University student
•Meal or Social Trip Purpose
PERSON CHARACTERISTICS
•High income
•Age over 50
•Female
•Preschool age
•Driving age HS student
BUILT ENVIRONMENT
•Mixed use origin/destination
•Intersection density
•CBD destination
•Residential density
BUILT ENVIRONMENT
•Rural origin
•Long walk/bike time
A Bit on Status
• All models estimated (around 50 or so.)
• Database design complete.
– Suzanne Childress will talk more about this.
• Software 2/3 complete.
– Jen Malm will talk more about this.