Dr.

Agnivesh, IIT (BHU) 09-04-2024

CE625 Traffic Engineering and Management

3.1-3.2

Lecture 1-2

Traffic Modal Split

(Step 3 after Trip Distribution)

Course Instructor
Indian Institute of
Department of
Technology (BHU)
Civil Engineering
Varanasi
Dr. Agnivesh Pani

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Introduction
• In a typical situation, a trip maker can
make choice among several travel
modes
• Personalized modes like car, two-
wheeler, bicycle, etc
• Shared public transit modes like bus,
train, metro, etc
• A mode choice or modal split model
is concerned with the trip maker’s
behaviour regarding the selection of
travel mode

CE625 Traffic Engineering and Management 2

CE621 1
Dr. Agnivesh, IIT (BHU) 09-04-2024

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Factors influencing Mode Choice

• Three broad categories of factors 3. Characteristics of
affecting mode choice: Transportation System
o Travel time
1. Socio-economic characteristics
o Travel cost
of trip-makers
o Reliability
o Household Income
o Comfort and convenience
o Car Ownership
o Family Size
o Age and Sex
2. Characteristics of the Trip
o Trip purpose
o Trip distance
o Time of the day

CE625 Traffic Engineering and Management 3

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Types of Modal Split Models

• Based on type of analysis used, models may be classified as:
1. Aggregate Model
Represents the average behaviour of a group of travellers
instead of a single individual (zonal-based)
2. Disaggregate Model
Represents the individual choice responses as a function of
characteristics of alternatives and socio-demographics of each
individual

CE625 Traffic Engineering and Management 4

CE621 2
Dr. Agnivesh, IIT (BHU) 09-04-2024

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Types of Modal Split Models

• Based on position of analysis used in the context of traffic
demand forecasting process, models may be classified as:
1. Trip-end Modal Split Model
2. Trip-Interchange Modal Split Model

CE625 Traffic Engineering and Management 5

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

• Used by early traffic engineers (may be called as early generation
models)
• Included both trip-end modal split models and trip-interchange
modal split models
• Trip-end modal split models used socio-economic characteristics of
trip makers as the primary determinants (why?) – A key variable used
in these models is car ownership
• Trip-interchange modal split models used service characteristics of
competing modes along with socio-economic characteristics of trip
makers

CE625 Traffic Engineering and Management 6

CE621 3
Dr. Agnivesh, IIT (BHU) 09-04-2024

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Southern Wiscosin Model
• Percentage of trip ends likely to use transit services from a zone was
related to (i) trip types, (ii) socioeconomic characteristics of the trip
makers and (iii) characteristics of the transportation system
• Trips were stratified as HBW, HBS, and HBO, and NHB Trips
• Average number of cars per household in a zone was considered as a
representation of socio-economic characteristics of trip-makers along
with number of households
• How to measure characteristics of transportation system?

CE625 Traffic Engineering and Management 7

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Southern Wiscosin Model
• Characteristics (or effect) of transportation system was considered in
terms of accessibility index of a zone:

CE625 Traffic Engineering and Management 8

CE621 4
Dr. Agnivesh, IIT (BHU) 09-04-2024

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Southern Wiscosin Model
• Relative transportation service provided by two modes (e.g., private
vehicle and transit) was characterized by the accessibility ratio:
𝐻𝑖𝑔ℎ𝑤𝑎𝑦 𝐴𝑐𝑐𝑒𝑠𝑠𝑖𝑏𝑖𝑙𝑖𝑡𝑦 𝐼𝑛𝑑𝑒𝑥
𝐴𝑐𝑐𝑒𝑠𝑠𝑖𝑏𝑖𝑙𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 =
𝑇𝑟𝑎𝑛𝑠𝑖𝑡 𝐴𝑐𝑐𝑒𝑠𝑠𝑖𝑏𝑖𝑙𝑖𝑡𝑦 𝐼𝑛𝑑𝑒𝑥
Example:

CE625 Traffic Engineering and Management 9

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Southern Wiscosin Model
Example:

CE625 Traffic Engineering and Management 10

CE621 5
Dr. Agnivesh, IIT (BHU) 09-04-2024

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Southern Wiscosin Model

CE625 Traffic Engineering and Management 11

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Diversion Curve Model
• Basic hypothesis underlying this model developed for Metropolitan
Toronto (Canada) was as follows:
 Total number of people moving between an origin-destination (O-D)
pair constitutes a travel market where various modes compete
Competing modes secure position as per their relative
competitiveness expressed in terms of relative travel time, relative
travel cost, relative transit service, and economic status of trip-maker

CE625 Traffic Engineering and Management 12

CE621 6
Dr. Agnivesh, IIT (BHU) 09-04-2024

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Diversion Curve Model
• Relative travel time is measured by travel time ration (TTR) between
competing modes 𝑇𝑇𝑅 =
• Where a = time spent in the transit
• b = transfer time between

CE625 Traffic Engineering and Management 13

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Diversion Curve Model
• Relative travel cost is measured by cost ratio (CR) between competing
modes 𝐶𝑅 =
( . )/
• Where i = transit fare

CE625 Traffic Engineering and Management 14

CE621 7
Dr. Agnivesh, IIT (BHU) 09-04-2024

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Diversion Curve Model
• Relative transit service is measured by the ratio of excel travel time by
transit and car
• Excess travel time was defined as the amount of time spent outside the
vehicle during a trip
• 𝑆𝑅 =

CE625 Traffic Engineering and Management 15

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Diversion Curve Model
• For a CR and different income groups, different diversion curves were
developed for different SR values

CE625 Traffic Engineering and Management 16

CE621 8
Dr. Agnivesh, IIT (BHU) 09-04-2024

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Aggregate Modal Split Models

Limitations
• Captive and choice transit riders were not identified and represented
separately in the models
• For this reason, models failed to reflect adequately the way choice
transit riders react to changes in transport system characteristics
• Data used were of zonally aggregated nature

CE625 Traffic Engineering and Management 17

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Disaggregate Mode Choice

Models

CE625 Traffic Engineering and Management 18

CE621 9
Dr. Agnivesh, IIT (BHU) 09-04-2024

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Advantages over Aggregate Models

• Disaggregate approach explains why an individual makes a
particular choice given the circumstances and is better able
to reflect changes in choice behaviour due to changes in
individual characteristics and/or attributes of alternatives
• Because of their causal nature, they are likely to be more
transferable to a different point in time and to a different
geographical context (a critical requirement for prediction)

CE625 Traffic Engineering and Management 19

20
Transportation – Choice Set

driving

Work

taking
transit

Home
walking/
bicycling

other

CE621 10
Dr. Agnivesh, IIT (BHU) 09-04-2024

Stated Preference ~ Revealed Preference

21

Characteristics Revealed Information Data Stated Information Data

Choice behavior in actual market Choice behavior in hypothetical scenarios

Preference
Cognitively incongruent with actual
Cognitively congruent with actual behavior
behavior
Actual alternative Generated alternative
Alternatives Responses to new or non-existing
Responses to existing alternatives
alternatives

May include measurement error No measurement error

Attributes Multi-collinearity can be avoided by

Correlated attribute
design
Ranges are limited Range can be extended

Choice Set Ambiguous in many cases Pre-specified

Responses Difficulty for multiple responses Repetitive questioning

Various responses formats(e.g., choose one,
Response Format Preference information available is choice
ranking, rating, matching)

Example of RP questionnaires design

22

CE621 11
Dr. Agnivesh, IIT (BHU) 09-04-2024

Example of SP questionnaires design

23

SP design – Choice cards

24

CE621 12
Dr. Agnivesh, IIT (BHU) 09-04-2024

Preferences - Discrete
25

 Discrete choice of one option from a set of competing ones. This response measures the
most preferred option relative to the remaining, but provides no information about the relative
preferences among the non-chosen. That is a true nominal scale.
Example-Auto > bus, train, ferry, carpool
bus = train = ferry = carpool

Mode for journey to work Consumer chooses

Bus
Train
Ferry
Auto ✓
Carpool

Preferences - Binary
26

 ‘Yes, I like this option’ ‘No, I don’t like this option’. This response clearly separates
alternatives into liked and not liked options and provides preferences.
Example- Auto > bus, train, ferry
 Carpool > bus, train, ferry

 Auto = carpool; bus = train = ferry

Mode for journey to work Consumer chooses (yes/no)

Bus No
Train No
Ferry No
Auto Yes
Carpool Yes

CE621 13
Dr. Agnivesh, IIT (BHU) 09-04-2024

Preferences - Ranking
27

 Complete ranking of options from most to least preferred. This response orders all options
on a preference continuum, but provides no information about degree of preference, no order.
 Example- Auto > bus, train, ferry, carpool
 Carpool > bus, train, ferry
 Ferry > bus, train
 Train > bus Mode for journey to work Consumer chooses

Bus 5
Train 4
Ferry 3
Auto 1
Carpool 2

Preferences – Rating (1-10)

28

 Rating options on a scale. Expresses degrees of preference for each option by rating them
on a scale or responding via other psychometric methods such as magnitude estimation. If the
consumers can supply valid and reliable estimates of their degree of preference this response
contains information about equality, order and degrees of differences and magnitude.

Mode for journey to work Consumer chooses

Bus 4

Train 4

Ferry 6

Auto 10

Carpool 7

CE621 14
Dr. Agnivesh, IIT (BHU) 09-04-2024

The Choice Modeling Process

29

Choice
set

Attributes of
alternatives
Parameters

Input Output
Mathematical Models Data
Data

Characteristics of
decision makers

Decision
makers

Utility-Based Choice Theory: Random Utility Models

If analyst understood all aspects of the

internal decision making process of decision Deterministic Choice Models
makers as well as their perception of
alternatives

Analysts do not have such knowledge

Analysts do not understand the decision

process of each individual or their perceptions
of alternatives

Random Utility Models

Analysts do not have full information about all
attributes of alternatives considered by the
decision makers

No realistic possibility of obtaining this

information
30

CE621 15
Dr. Agnivesh, IIT (BHU) 09-04-2024

Elements of Choice Decision Process: Attributes of

Alternatives
31

 Alternatives in a choice process are characterized by a set of attribute

values
 Generic attributes: Apply to all alternatives equally
 Alternative-specific attributes: Apply to one or a subset of alternatives
 Wait time at a transit stop or transfer time at a transit transfer point
are relevant only to the transit modes
 Attractiveness of an alternative is determined by the value of its attributes
 Measure of uncertainty about an attribute can also be included as part of
the attribute vector in addition to the attribute itself
 For example, if travel time by transit is not fixed, the expected value of
transit travel time and a measure of uncertainty of the transit travel time
can both be included as attributes of transit
 Important to identify policy-related attributes!
 Measure of services (travel time, frequency, reliability of service, etc.) and
travel cost

Utility-Based Choice Theory:

Basics of Utility Theory
32

 Utility is an indicator of value to an individual.

 Utility maximization rule states that an individual will select the alternative from his/her set of
available alternatives that maximizes his or her utility

 The rule implies that there is a function containing attributes of alternatives and characteristics of
individuals that describes an individual’s utility valuation for each alternative

 Alternative, ‘i ’, is chosen among a set of alternatives, if and only if the utility of alternative, ‘i ’, is
greater than or equal to the utility of all alternatives, ‘j ’, in the choice set, C.

U ( X i , Sn )  U ( X j , S n )j  i  jj  C
Utility
function Vector of attributes
describing alternatives i and j Vector of characteristics
describing individual n

CE621 16
Dr. Agnivesh, IIT (BHU) 09-04-2024

Probabilistic Choice Theory: Random Utility Approach

33

 The individual is assumed to choose an alternative if its utility is greater than that of any other
alternative

 The probability prediction of the analyst results from differences between the estimated utility
values and the utility values used by the decision maker

 How to represent this difference?

 Decompose the utility of alternative!

Portion of the utility

Portion of the utility observed unknown to the analyst
by the analyst “Deterministic “Random Error Term”
Portion of the Utility”

U in  Vin   in

Components of the Deterministic Portion of the Utility

34

 “Deterministic -- Observable -- Systematic” portion of the utility!

 Mathematical function of the attributes of the alternative and the characteristics of the decision
maker

 Any mathematical form but generally additive to simplify the estimation

Systematic portion of
the utility for alternative
i for individual n; Vin
Vin  V ( Sn )  V ( X i )  V ( Sn , X i )

Characteristics
of decision Interactions between the
maker n Attributes of attributes of alternative i
alternative i and the characteristics of
decision maker n

CE621 17
Dr. Agnivesh, IIT (BHU) 09-04-2024

Mode Choice Modeling

35

Travel as a choice process

36

CE621 18
Dr. Agnivesh, IIT (BHU) 09-04-2024

Two stage modal split model

37

All Trip

Non-motorized Motorized

Bicycle trip Public Transit Private Vehicle

Walk trip

Bus Transit Rail Transit Car Two-wheeler

Normal Bus Rapid Bus Metro Suburban Rail

Factors that Affect Mode Choice

38

 Trip characteristic
 Trip distance (short trip or long trip)
 Time of day of travel; trip purpose
 Trip maker’s characteristic
 Family income
 Car/vehicle ownership
 Family size and composition
 Residential density
 Transportation system’s characteristic
 Travel time; waiting time; Travel cost
 Comfort and convenience
 Reliability and regularity; Protection and security
 Accessibility
 Travel time ratio (TTR)
 Travel cost ratio (TCR)

CE621 19
Dr. Agnivesh, IIT (BHU) 09-04-2024

Travel time
39

𝑇𝑇 =𝑡 +𝑡 +𝑡 +𝑡 +𝑡

Where,
𝑇𝑇 = Travel time for public transit
𝑡 = Walking time from origin to public transit
𝑡 = Waiting time for public transport
𝑡 = In – vehicle time for public transport
𝑡 = Time spent in transfer from one PT to another PT
𝑡 = Walking time from PT to destination
𝑇𝑇 =𝑡 +𝑡 +𝑡

Travel Time
40

 Travel time ratio – TTR

𝑇𝑇 𝑡 +𝑡 +𝑡 +𝑡 +𝑡
𝑇𝑇𝑅 = =
𝑇𝑇 𝑡 +𝑡 +𝑡

𝑡 =𝑡 ;𝑡 =𝑡 ;𝑡 =𝑡 ;𝑡 =𝑡
𝑡 =𝑡 ;𝑡 =𝑡 6
𝑡 =𝑡 ;𝑡 =𝑡

 Travel cost ratio – TCR

𝐶𝑜𝑠𝑡 𝑜𝑓 𝑡𝑟𝑎𝑣𝑒𝑙 𝑏𝑦 𝑝𝑢𝑏𝑙𝑖𝑐 𝑡𝑟𝑎𝑛𝑠𝑖𝑡
𝑇𝐶𝑅 =
𝐶𝑜𝑠𝑡 𝑜𝑓 𝑡𝑟𝑎𝑣𝑒𝑙 𝑏𝑦 𝑐𝑎𝑟

CE621 20
Dr. Agnivesh, IIT (BHU) 09-04-2024

Utility/Disutility
41

𝑈 = 𝛽 + 𝛽 𝑥 +𝛽 𝑥 +𝛽 𝑥 +……….+𝛽 𝑥
𝑈 = 0.001 − 0.0183 𝑇𝑇 − 0.0139 TC − 0.0151 DC
𝑈 = −0.6255 𝑊𝑇 − 0.5569 TT − 1.665 𝑇𝐶 − 3.124 𝐷𝐶
𝑈 = −0.3952 𝑊𝑇 − 0.3346 TT −1.572 𝑇𝐶 − 2.365 𝐷𝐶

 Mode specific utility models

 Abstract utility models
Generic variables
𝑈 = 𝛽 − 0.165 𝑇𝑇 − 0.125 TC − 0.018 WT - 0.152 DC

Multinomial Logit Model (MNL)

42

CE621 21
Dr. Agnivesh, IIT (BHU) 09-04-2024

Examples
43

©SCULPT{Lab}, IIT (BHU) Varanasi (https://www.sculptlab.in/) Dr. Agnivesh Pani

Thank You!
“Forty hour workweeks are a relic of the Industrial Age. Knowledge
workers function like athletes — train and sprint, then rest and
reassess.”
― Naval Ravikant

CE625 Traffic Engineering and Management 44

CE621 22