Transportation Planning

(BTL2034)

By: Dr. Seyed Mohammadreza Ghadiri , April 2021

 Problem Definition

Let us first pose a general question as follows:

What is the minimum level of services of transport system desired by the

urban community to improve their quality of life in the future?

The overall objective of community is to improve their quality of life in

the future that is the expected goal of any urban community. So, if you
can answer this question in very simple terms probably, we can
understand the three important terms related to problem definition phase
namely, goal, objective and standard.

Question 1: what kind of system they expect to improve their quality of

life?
 Problem Definition

The simple answers to that question are as following:

a. Journey speed the public transport to be not less than 25 kilometres per
hour,
b. The frequency of services of public transport on any route to be not less than 3
per hour
c. The occupancy of public transport vehicles not to exceed the permissible limit
even during peak hours. Everyone, should be able to get into the vehicle
whenever, they want to travel by transit vehicle.

These statements are nothing but the standards about which we already
discussed in details. Standards are derived from objective. Objectives are
based from the community goal right. So, that is how we need to understand
the terms goal, objective and standards in just answering a simple question
related to the desired of urban community to improve their quality of life in
the future.
 Problem Definition

Question 2: What are the constraints in achieving the desired level of service
of transport system in the future?
a. Financial constraint: The total possible investment for augmentation
of transport infrastructure for example will be about US dollars 5
billion in the horizon year which is the possible investment from all
sources.
❑ So, this is the financial constraints if this constrains known to the planner.
Planner can definitely orient the planning process that this constraint is
also taken care of.
b. Political Constraint: Government might say that the maximum
possible subsidy to public transport services whether train or bus will
be above US$ 500 million per annum - that is the maximum possible
subsidy the government can provide.
❑ So, if this information is available to the planner then this constraint can be
put into the planning process and while analysing the system
characteristics.
 Problem Definition

It is understanding the cost implication of transport system after taking into

account the subsidy accepts. So all these statements are related to the
component constrained in problem definition phase.

Question 3: What are the factors that may influence the future demand for
transportation in the urban area?
We are concerned about these factors because the planner should know what
the total travel demand is in the horizon year or in the future for which the
system has to be planned. To quantify the travel demand we should understand
the factors influence the demand for transportation and how these factors are
going to change compare to today’s situation.
a. Population growth
❑ To answer this question, the planner should be clearly informed that there is
going to be population growth in that particular urban area since there are
cities around the world where population is even decreasing!
 Problem Definition

b. Decrease of household size

c. Increase of household income
d. Increase of household vehicle ownership
e. Increase of per capita trip rate

So, these should be given as input to quantify the total travel demand in the
future scenario all this things are related to which component in the problem
definition phase. Inputs are nothing but the demand generated by the social
economic characteristics of the urban dwellers. So, once you are able to answer
the question you can understand the definition and implication of components
inputs.
 Problem Definition

Question 4: What are the quality aspects of transport system that may
influence the future demand? Keep in your mind that this question is related
to outputs of a system.

o Speed of travel
o Travel cost
o Level of safety
o Level of comfort
 Problem Definition

Question 5: How to assess the quality of future transport systems?

o The answer is this estimate money value of the relevant system outputs
like speed, level of safety, and so on using appropriate functional
forms.
Question 6: How to choose the system that will provide the better quality of
services? It means when we have several alternatives, how do we choose the
system that will provide better quality of services?
o This question is related to decision criterion. So, what is the criterion?
Use cost-benefit analysis and stated preference approach to choose the
optimal system.

❑ If this information is provided to the planner beforehand than he or she is

very clear to how to proceed with the planning process. That is the idea of
defining the problem with all the six important components. So, with this
understanding, let us go further in the systems engineering process.
 The Systems – Engineering Process

You can realize now, that we have completed the decision on the details listed
in the first two boxes only so far, and next step is solutions generation. Then
solutions analysis evaluation and choice, recommend strategy and
implementation.
Socioeconomic Environment

Problem Definition (i) Objectives,

(ii) Constraints (iii) Inputs, (iv)
Implementation
Outputs, (v) Value Functions & (vi)
Decision Criterion

Recommended Strategy
Solution Generation

Solution Analysis

Evaluation & Choice

 Solution Generation

Question: What do you mean by solution generation?

o The term implies generation of different alternative solutions to the
transportation problem and the aim of this phase is to generate an array
of solutions that satisfies the previously established objectives to a
lesser or greater degree and which does not violate the constraints.
 Solution analysis

The objective of this step is to predict the probable operating state in terms of
speed, comfort, right cost and so on of each of the alternative systems
generated in the previous phase, given expectations about the state of the
environment, state of the social economic characteristics and land use pattern in
the horizon year, and this is normally done through a systematic analytical
procedure.

o In urban transport planning, the input magnitudes (which means the

demand for transportation inputs), and the behaviour of alternative
systems are estimated normally through the use of a four-phase process
consisting of trip-generation analysis, modal split analysis, trip-
distribution analysis and traffic-assignment analysis.
 Evaluation and Choice

The aim of this step is to identify the alternative system that satisfies the
objectives to the greatest extent.

o The data required for this step are the output variable magnitudes
predicted in the system analysis step, the value functions, and the
decision criterion identified in the problem definition step.
 Implementation

The optimum system for the horizon year is identified in the previous phase
and the aim of this phase is to formulate a strategy for implementing the
chosen system throughout the planning period.

❑ As we said earlier strategy is normally facing the project

implementation which means implementing the projects in phases. We
fix the targets that are what should be completed on fifth year, tenth
year, fifteenth year, twenty year, and so on and carry out the
development as per the plan strategy.
 The Systems – Engineering Process

Now we know clearly what do we mean by each of the steps in system

engineering process, which can also be directly applied for transportation
system scanning process.

Socioeconomic Environment

Problem Definition (i) Objectives,

(ii) Constraints (iii) Inputs, (iv)
Implementation
Outputs, (v) Value Functions & (vi)
Decision Criterion

Recommended Strategy
Solution Generation

Solution Analysis

Evaluation & Choice

 Definition of Urban Area

An urban area is the region surrounding a city. Most inhabitants of urban areas
have non-agricultural jobs. Urban areas are very developed, meaning that there
is a density of human structures such as houses, commercial buildings, roads,
bridges, and railways. "Urban area" can refer to towns, cities, and suburbs.

❑ An urban area is a location characterized by high human population

density and vast human-built features in comparison to the areas
surrounding it. Urban areas may be cities, towns or conurbations, but
the term is not commonly extended to rural settlements such as
villages and hamlets. (Wikipedia, 2015)
 Definition of Urban Area (Ref: www.unicef.org, 2015)

❑ The definition of ‘urban’ varies from country to country, and, with periodic
reclassification, can also vary within one country over time, making direct
comparisons difficult. An urban area can be defined by one or more of the
following:

❑ Administrative criteria or political boundaries (e.g., area within the

jurisdiction of a municipality or town committee), a threshold population
size (where the minimum for an urban settlement is typically in the region of
2,000 people, although this varies globally between 200 and 50,000),
population density, economic function (e.g., where a significant majority
of the population is not primarily engaged in agriculture, or where there is
surplus employment) or the presence of urban characteristics (e.g., paved
streets, electric lighting, sewerage).
▪ 68% of the world population projected to live in urban areas by 2050,
says UN
https://www.un.org/development/desa/en/news/populatio
n/2018-revision-of-world-urbanization-prospects.html
 Demand for Transportation

Question: How do we quantify the demand for transportation in the future

year?

❑ First of all you must have some frame works. For example if you want to
analyse the demand at the micro level we should understand that trips
made by the urban travellers emanate at home (from home they travel to
different places and come back in the evening). In certain cases trips
might emanate from work places shopping areas recreational centres and
so on.
❑ If you wish to analyse trips generated by each of the households, each of
the residential building in urban area and if you collect data about the
trips made by each of the households then it will be highly complex data
set which is not that much easily to analyse.
 Demand for Transportation

❑ There is no need also to go to that micro level to plan for transportation

system for an urban area. We can always think of aggregating the points of
trip generation and trip attraction so that we have a manageable amount of
database which can be analysed with clear understanding the results and
which can be used to plan for the appropriate transportation system.

Question: How to aggregate and analyse the travel pattern?

❑ Instead of considering each household (as a point of origin or

destination) why not just mark of small area of more or less similar type
of land use of radius say about 1kilometer radius or one and a half radius
even 2 kilometre radius depending upon the homogeneity of land use,
and consider that smaller area as a single unit and quantify the total trips
produced or attracted by that area. Similarly divide the entire urban space
in the smaller land use parcels and treat each of these land use parcels as
points of trip generation.
 Demand for Transportation

As the first step we need to delineate the urban space is to fix the boundary
for our study area. This boundary normally extends beyond the municipal
boundary of any town or city because normally people living in peripheral
areas might also travel on regular and daily bases into the city and go back
home. So, you must study the travel pattern very carefully and cover all the
points of origin and destination which contribute for the regular travel pattern
on daily bases in the urban area that should be the bases for fixing the
boundary.

❑ Then divide the area into smaller parts. For now you just assume that it is
possible to divide the urban area into smaller land use parcels based on
certain conditions.
 Demand for Transportation

Then we call this land use parcels as land use parcel or traffic zone, because we
are going to assume that traffic is going to generate from this smaller land use
parcels. Another important aspect is fixing 1 single point for each of these small
land use parcels which is assumed to be the trip generating point. We are going
to assume that all the trips generated from particular origin, then analysis of
travel pattern is easier.

❑ What we are going to do from now on is just to study the movement pattern
or mobility pattern between all these identified points that is a point one to
all other points, point 2 to all other points and so on.

❑ So, that way it will be easy for us to understand the travel pattern and
analyse the total transportation demand. This is the first important step and
then once we accept this process than we get into the analysis phase.
Transportation Statistics in Malaysia

https://www.tourism.gov.my/statistics
 Transportation Statistics in Malaysia

https://www.malaymail.com/news/malaysia/2019/12/13/malaysia-forecasts-
slow-travel-growing-trend-next-year-says-deputy-minister/1819061
 Travel Demand Forecasting

❑ Travel Demand
Number of persons/vehicles expected to travel between a particular
origin and destination via a particular route and mode of travel over a
given time.

▪ Required for:
• New construction
• Expansion of existing facilities
 Four-Step Travel Demand Model

❑ Originally developed in 1950’s and 1960’s when planning major

highway facilities. Four decisions are the basis of the traditional
travel demand model:

▪ The choice and reason to travel

▪ The destination to travel to
▪ The mode by which to travel
▪ The route on which to travel
 Four-Step Travel Demand Model

1. Trip Generation
▪ How many trips will be generated by a given location and when will
these trips occur?
2. Trip Distribution
▪ What are the origins and destinations of these trips?
3. Mode Choice
▪ Which mode of transportation will be used to make the trip?
4. Traffic Assignment
▪ Which route on the transportation network will be used when making
the trip?
 Four-Step Travel Demand Model

❑ Example of steps in the Four-Step Planning Process

 Travel Demand Forecasting

❑ Delineate the study area boundaries into traffic analysis zones (TAZ)

TAZ is defined as:

▪ “a special area delineated by
state and/or local
transportation officials for
tabulating traffic-related data
especially journey-to-work
and place-of-work statistics”.
(US Census Bureau)
 Demand for Transportation

Land use parcel or traffic zone

 What Factors Influence Travel Demand?

❑ Location and intensity of land use

❑ Extent, cost and quality of available transportation services
❑ Socioeconomic characteristics of study area
Question: Where to get this information? Answer: National Census (every 10 years), Travel Diaries or surveys
https://nhts.ornl.gov/2009/pub/fielddocuments.pdf
What Factors Influence Travel Demand?
What Factors Influence Travel Demand?
 Solution Analysis

There are other steps involved which are quite closely related to this 4 steps
let us take the overview of the whole analytical process including the data
collection process in the form of flow chart.

❑ The next flow chart gives you complete picture of data collection analysis
and travel demand prediction for the horizon year. The analysis is divided
into 2 parts 1 part pertains to base year condition the other part pertains to
horizon year condition. An inter connectivity between the steps are shown
by arrows.
 Sequence of Activities in Transport Analysis

Base Year Horizon Year

Land-use Survey Land-use Allocation

Alternative Transport System

Travel Survey

Trip Generation Models Zonal Trip Ends

Captive Modal- Split Models

Choice Modal- Split Models

Trip Ends by Mode
Trip Distribution Models
O-D Volumes by Mode
Route Assignment by Model Models
Route Assignment by Mode
Check Against Measured Volumes
Volumes by Link and Mode
 Solution Analysis

❑ The first important step in the base year condition is land use survey. We
know what are the categories of urban land use. We must collect data with
regard to type and intensity of use in each type of the land use over the
entire urban area.

❑ You must have all the information related to travel pattern of the urban
dwellers. This includes the related traffic service also concerning the way
the vehicles the actual traffic experienced the terminals the fleets price of
the traffic system and so on. All the information will be collected while we
do travel survey.

❑ Then this will go as input to the 4 important analytical steps namely trip
generation, modelling (the models split into 2 stages captive models-
split models, choice models-split models), then trips distribution models,
route assignment then check against measured volumes.
 Solution Analysis

Question: What exactly we are going to do in trip generation analysis?

Trip generation has got two aspects named trip production, and trip
attraction. Trip production and trip attraction are related to estimation of the
trip produced and attracted by each of the traffic zones that we have identified
earlier. This is zone wise estimation of the total trip production and attraction.
We will have some number of total trip produced and trips attracted for each
of zones.

❑ We will just try to relate the trip production with this set of causal factors.
So, that is how we get some equation to quantify the trip production, and
the trip attraction.
 Solution Analysis

Once you quantify the total trips we need to know the mode chosen by the
travellers at each of the traffic zones. It will be convenient for us first if we
segregate the captive travellers.

Question: Who are the captive travellers?

The travellers who are captive to particular mode are captive travellers. Let
us say there are certain percentage of households with no vehicles not even
bicycle then, what will be their mode of transportation. What will be their
vehicle? Mode of transportation is public transport only. So, there is no
choice available to this section of the travellers.
 Solution Analysis

Question: Why we include this section in mode choice analysis when they
have no choice?
You can segregate that portion of the travellers from your mode choice
analysis and this process is what we called here as captive model split
model. It is not really modelling, it is a simple analytical producer of
segregating captive travellers from the whole of the travellers then we will
have later pertaining to choice travellers only who have real choice than
develop a model to understand the mode choice process identify the factors
that will influence choice of mode of travellers then just develop an
equation, which will give an idea about probability of choice of the
particular mode by the particular type of the traveller and use this equation
to get actual mode choice at each of the traffic zones.
 Solution Analysis

We identified already in the previous step we have estimated the total trip
production and attraction, after compilation of mode choice analysis. We will
know how many are, what percentage will take transit, what percentage will
take motorized two wheeler, what percentage will take car, what percentage
will go by foot all the information will be available again zone wise. That is
what will be doing in mode choice.

❑ Now, we look at the interchange of trips between the zone centroids (the
points that we are already fixed) and try to identify first the factors that
will influences the trip distribution.
 Solution Analysis

Question: What are the possible factors that might influence the distribution
of trips? How people travel from one zone to another zone?
❑ One possibility is their home is located in one zone work place is in
another zone. So, they have to travel to that zone to go to work or if there
are children going to school it might be located in different zone and they
have to travel to different zone. A shopping mall will be available in some
other zone and they travel one zone to another zone. So, identify the
causal factors, trip purpose as well as other related issues.

❑ Normally, when the distance of travel involves from a particular zone is

more as a distance increases number of trip exchanged will be decreasing.
Normally urban dwellers would like to minimize the intensity of travel
into different activities. So, all these information will be analysed and will
develop a model to explain the trip distribution process.
 Solution Analysis

❑ So, we will distribute the trips as person trips not in terms of vehicular
trips even in the distribution process we distribute the trip as person trips.
We can distribute the trips using an analytical technical and then check
cross with the data that will be collecting during the base year condition
then in the fourth step which is route assignment. We actually convert the
person trips into equivalent vehicular trips.

❑ We must understand the choice of the route by the travellers. There will be
some factors influencing the choice of route. We must identify the factor
and then try to develop the equation or model to explain the route choice
process and then the converted a vehicular traffic as to be assigned along
various routes bases on our calculation. Once you do that you actually
reproducing the actual traffic flow by theoretical means.
 Solution Analysis

In short, we started with quantifying the total trip production and trip
attraction at the zonal level then understanding the choice of mode then
understanding distribution of trips on space initially between zone centroids,
then assigning routes for these trips after converting the trips into equivalent
vehicular trips. Once you do these exercises if you know the road geometry of
the entire road network on each link then you will be able to visualize it
theoretically.

❑ What will be the traffic volume over the entire road network after traffic
assignment?
▪ You should get information about traffic volume as well as composition.
 Solution Analysis

Another important aspect to be kept in mind in traffic assignment is that

assignment of traffic by transit is easier because transit services are operated
on fixed routes. So, once you know the origin and destination of the
passengers you can easily understand the route which will be taken by the
traveller and you can put all this kinds of travellers into that particular route.

❑ Transit assignment is simpler and assignment is more complex in the case

of private or personal vehicle uses. Assignment has to be done in detail
only for personal vehicles users. Transit is quite simple because route is
already fixed. Once you assign the personal vehicle on the road along with
the transit service may be the bus service we should be able to visualize
the total traffic flow on the road network or we should know the total
traffic volume as well as composition at any point of time in urban road
network.
 Solution Analysis

❑ You must take care of every step to see that your theoretical model is
valid. You can cross check statistically logically and so on and ensure that
at every stage your equation or model is reproducing the reality.

❑ Once you are able to theoretically reproduce the actual traffic flow you
have the advantage of comparing the theoretically productive traffic flow
with the actual observed traffic flow for the base year condition.
 Solution Analysis

If the two values are matching you can be happy and your model is perfect. If
not you have to go back work back and look for possible errors correct
wherever required and then rework on the four steps and get again the revised
traffic volume compare with the actual traffic volume and iteratively do this
exercises until there is the reasonable match between theoretically estimated
traffic volume and filed observed traffic volume.

❑ If it is matching then your model is able to explain the reality a reasonable

extend.
 Solution Analysis

Question: Why exerting ourselves to develop this kind of model? What is

the purpose?
It seems we can measure trip production, mode choice trip distribution, and
route assignment for base a condition. Once you fix the zone boundary you
can just actually conduct services and get the trip production value, trip
attraction value, and so on. So, why should we develop a model?

❑ Because we need to have some tool to predict the future trip production,
future trip attraction, future mode choice, future trip distribution, and
future traffic assignment that is a ultimate objective.

❑ That is why we are interested to explain what is happening today

theoretically. So, these theoretical models can be used to predict the future
scenario as accurately as possible. That is what is shown on the right hand
side of previous flow chart.
 Solution Analysis

Question: In this context what we do for the horizon year?

We predict the land use allocation, which involves prediction all the
population growth, household changes and land use changes everything
together. The future urban land use is predicted.
 Solution Analysis

There are models available about to predict the future land use pattern of an
urban area. Once you do that you just use this model trip generation, model
use the land use data and get information about zonal trip ends.

❑ Traffic zones will be there for future condition also if the area is going
to expand extend your zoning process and fix a points of trip origin
destination using the models you can find out the total trip ends
produced and attracted at each of the traffic zones.

❑ So, once you use this model for the future condition you get zonal trip
ends at of the traffic zones. Then use the captive mode choice model and
choice mode choice model and if you are going to introduce new modes
that will become into the choice process.
 Solution Analysis

❑ Introduce that also for the horizon year and user mode choice model to
analyse the mode choice process for the horizon year condition and get the
trip ends by mode. If you’re going to introduce metro rail will get another
new mode, but still your model is valued to analyse such situation also it is
possible.

❑ So, you will get trips ends by mode for each of the traffic zones for a
horizon year condition then use your trip distribution model and understand
the O-D volumes by the mode for horizon year condition trip distribution
for the horizon year condition. Then route assignment can be done directly
from the O-D volumes.
 Solution Analysis

❑ There is no input required further for the horizon year condition and once
assigned route you get volumes by link and mode for the future
transportation network that is what we required.
❑ Once you get this volume we can check whether the available
infrastructure is adequate or not. You may have a limited width of road, but
after assignment of the horizon year condition you may get the double the
capacity of existing road way, that means definitely you need to widen that
particular stretch of road. If it is a continuous set of links along a route that
is going to be major traffic corridor.

❑ If that is the situation than you can think of introducing new systems like
light train metro rail system and so on one it is possible by this process to
identify major traffic corridors and decide about the transport system it will
suit your future requirement.
 Solution Analysis

❑ So, this is what do we need to understand by the sequential activities in

transport analysis process.
❑ With this, we have developed some understanding about the problem
definition process and the major analytical steps involved in development
of demand models to ultimately get a future travel demand and required
augmentation for a transportation infrastructure.
 Recapture!

❑ So far we were discussing about the different steps in the system

engineering process. To be more specific we discussed about solution
generation phase, solution analysis, evaluation and choice, and
implementation in connection with system engineering process which are
directly applicable for transport system planning process.

❑ Then we understood that it is necessary to quantify the inputs and the

outputs related to a system more accurately, and more important to quantify
for the horizon year condition.

❑ The system inputs are nothing but, the quantum of demand for
transportation in future year, and system outputs are nothing but, the system
characteristics that you are planning for meeting the demand in the horizon
year. so we need to do detail analysis to quantify the input and then to
quantify the characteristics of the system that we are purposing.
 Recapture!

❑ In that regard we understood that there are four important analytical steps
followed to get the answer for this questions namely:

▪ What would be the total demand in the horizon year? and

▪ What would the characteristics of the system that we purposed for

the horizon year condition.
 Sequence of Activities in Transport Analysis

Base Year Horizon Year

Land-use Survey Land-use Allocation

Alternative Transport System

Travel Survey

Trip Generation Models Zonal Trip Ends

Captive Modal- Split Models

Choice Modal- Split Models

Trip Ends by Mode
Trip Distribution Models
O-D Volumes by Mode
Route Assignment by Model Models
Route Assignment by Mode
Check Against Measured Volumes
Volumes by Link and Mode
 Trip Generation Analysis

Question: What does trip generation analysis mean?

▪ It is understanding of the trip generated in different land use parcels.

❑ To define little more precisely, we need to understand what exactly mean

the term trip? We normally understand the trip as information related to a
journey. We ask a friend how was the trip? This means all aspects related to
the journey.

Question: What does trip mean?

A trip for the purpose of analysis is defined as a one-way movement from
an origin to a destination.

Origin Destination
 Trip Generation Analysis

Question: For the purpose of analysis how do we define origin in connection

with a trip and what is the definition of destination?
▪ We divide entire urban area into smaller land use parcels named traffic zone
and then will fix the points of origin as well as destination as zones
centroid, so what the mean here by trip is movement from one point to
another point from one zone centroid to another zone centroid.
 Trip Generation Analysis

❑ The principle task of trip generation analysis is to relate the intensity of trip
making (number of trips made from one point to several other points) to and
from land use parcels to measures of the type and intensity of land use.

❑ There are two types of trip generation analysis and these are trip
production and trip attraction.

▪ Trip Production: the term trip production refers to the trips generated
by residential zones where these trips may be trip origins or trip
destinations. So these are trip associated with residential zones.

▪ For example, a trip made from home to work is a home based of trip
production, a trip made from workplace to home is also a trip
production, a trip made from home to school trip is also production,
school to home again trip production because one end of the trip is
associated with home.
 Trip Generation Analysis

▪ Trip Attraction: The term trip attraction is used to describe trips

generated by activities at the non-home end of trips.

▪ This term is used to describe trip generated at the non-home end, for
example it could be workplace to shopping area, shopping area to
workplace or one workplace to another workplace, so all these
movements are trip attractions.

❑ Of course, urban travel demands are made up of a number of different trip

types that have specific spatial and temporal characteristics.
 Trip Generation Analysis

We classify the trips, at least broadly, to go further in the analytical process.

the first level of trip classification use normally is a broad grouping into
homebased trips and non-homebased trips.

❑ Home based trips, and

❑ Non-homebased trips

o We analysis home based trip separately and non-homebased trips

separately. We cannot combine both this categories of trips and develop
models, we need to develop separate trip generation models for home-
based and non-homebased trips because the type of variables that might
influence the trips are going to be different or even if the variables types
are same the extent of influence of this variable on this trip making might
be different.
 Trip Generation Analysis

❑ Homebased trips are those trips that have one trip end at a household
(either the origin or the destination).

❑ Examples of non-homebased trips are trips between workplace and

shopping area; business trips between two activity centers, etc.
 Trip Generation Analysis

Question: What is household in the context of trip generation analysis?

▪ Previously we assumed household to be just a family here we know
that household encompasses a different sets of people including a
family. A household is usually a group of persons, who normally live
together and take meals from a common kitchen unless the
exigencies of work prevents any of them from doing so.

o NOTE: One person of the family may have to leave for work very
early in the morning 5 o'clock and come home may be very late
around 11 o'clock. That case he/she may not be able to have food
from the same kitchen but, still this person can also be considered to
be a member of the household.

❑ Person in a household may be related or unrelated to one another or a mix

of both. It means that persons in a household need not to be related.
 Trip Generation Analysis

❑ A group of unrelated persons who live in an institution and take their meals
from a common kitchen is called an institution household.

Examples: Hostels, rescue home, orphanages, jails, hotels, etc.

▪ It is very important to understand the term institutional household

because later we may have to collect data for transportation planning
process from households.
▪ If we understand that households are only people who are living in
houses, who are mostly related to one and other, then we will be
missing out other categories of households. So it is very important we
consider the normal households as well as institutional households.
 Trip Generation Analysis

1 a
▪ These are three land use
parcels on traffic zones
R W
numbered as 1, 2 and 3. b
Let’s assume W stands for
workplace (zone 2 is g c 2
completely work zone and
no other land use activity is h d
there just a workplace, may
be a factory location or e
location with full of
institutional and offices) and
f
R stands for residential area,
and S stands for shopping.
S
3 R
❑ Zone 3 is partly residential and partly commercial this is the reality you may
have mix land uses in traffic zones and it may be undividable.
 Trip Generation Analysis

1 a
R W
a , b – home based trip b
c , d – non-homebased trips 2
g c
e , f – home based trip
g , h – home based trip h d
e
▪ A total of 8 trips are involved in
f
this particular area. S
3 R
 Trip Generation Analysis

4 trip ends produced at zone one. Irrespective of the direction of movement

you just count the number of trip ends connected to the particular zone. If the
zone is a residential zone all are production so that is how we say that four
trip ends up produced at zone 1.

❑ Six trip ends are attracted to zone 2 because zone 2 is a workplace zone so
all the trip ends associated with non-home kind of activity is obviously a
trip attraction.

❑ 2 trip ends are produced at zone 3, and 4 trips attracted to zone 3. You
must look at the trip ends associated with R these are productions.
 Trip Generation Analysis

❑ The classification that have been used in major urban transport planning
studies for homebased trips are:

(i) Work trips, (mandatory)

(ii) Educational trips, (mandatory)
(iii) Shopping trips, (random)
(iv) Personal business trips
(v) Work related business trips, and
(vi) Social and recreational trips
 Modelling Trip Production

❑ The process of relating the trips produced by households to the factors

influencing trip production by appropriate analytical technique is termed as
trip production modeling.

▪ The number of trips produced depends on several factors related to the

household.
 Modelling Trip Production

❑ A large number of transportation studies have shown that the following

variables represent the most important households' characteristics related
to trip production:

(i) Household size and composition

(ii) Number of employed persons
(iii) Number of students
(iv) Household income
(v) Vehicle ownership, etc.
 Modelling Trip Production

❑ Transport planner has to collect this data from the households while
choosing variables for trip production modeling. You should be aware of
the amount of work involved in collecting data as well as the level of
accuracy, for example, if you try to collect data about household income
do you think that you will be able to get accurate estimate household
income? Even though it is very important factor influencing the trip
production rate most cases it may not be possible then what we do?

▪ We look for information which are proxies for household income e.g.
the vehicle ownership or other accessories the households have in the
household. So we cannot collect the household income directly but,
we can look for indirect indicators.
 Modelling Trip Production

Modeling trip production, in essence, involves relating a dependent variable

(trip production) to a set of independent variables (factors representing
household characteristics).

Question: What kind of analytical techniques we used for this kind of

relationship

▪ The most common analytical tool used for this purpose is the
regression analysis.
 Recapture!

❑ To recall our previous lecture you may remember we first try to define the
term trip. Trip is one-way movement from an origin to a destination.
❑ We also made a particular point very clear in connection with trip, as far as
origins of trips are concerned, we consider zones centroids as trip origins.
The destinations are also zone centroids. So, it is a just movement between
zones’ centroids.
❑ Then we tried to differentiate between trip production and trip attraction;
trips generated at residential zones are termed as trip productions and
trips generated at non-residential zones are trip attractions, then we
further classify the trips into home based and non-homebased trips.
 Recapture!

Question: Define home based and no-home based trips?

Answer: A trip having at least one end at household is a home based trip.
Obviously, non-homebased trips will have neither of its ends at
home, and the ends will be connected to non-home activity
centres.
 Recapture!

❑ We were trying to list the factors that might influence the production rate
of home based trips, and all the factors are related to household
characteristics, because households only mainly produce all the trips. Then
while modelling trip production we understood clearly that it is necessary to
classify the home based trips as work trips educational trips and so on.
❑ Because the set of variables that might influence these categories or these
types of home based trips could be different so their extended influence
trips also could be different.
 Recapture!

Question: What are the factors that might influence trip production?
Answer: Of course, the factors are related to household characteristics
including the size and composition of household, number of
workers in a household, number of students in a household,
household vehicle ownership, household income and so on.
 Recapture!

❑ We found that trip production analysis is nothing but that relating a

dependent variable to a set of independent variables.
❑ The dependent variable is the trip production rate for number of trip
produced by household and independent variables are related to household
characteristics the analytical tool used for this analysis is regression
analysis.

❑ We will now take up a very simple case of regression of relating one

independent variable with a dependent variable and just try to understand
the basic steps involved in the regression analysis, and finally, a static
namely coefficient of determination which will give some idea about the
effectiveness of the regression process .
 Basic Information About Regression Process

❑ Single Independent-Variable Case:

Assume that observations of the magnitude of a dependent variable Y have

been obtained for n magnitudes of an independent variable X and that an
equation of the form Ye = a + bX is to be fitted to the data.

Where,
Ye = Predicted scores (estimated values) of the dependent variable, Y
a = Intercept constant, and
b = Regression coefficient

▪ The above is the basic equation of single linear regression.

 Basic Information About Regression Process

❑ Single Independent-Variable Case:

https://www.youtube.com/watch?v=zPG4NjIkCjc&list=PLF596A4043DBEAE9C
 Basic Information About Regression Process

❑ For the least-squares criterion, the magnitude of b can be calculated as:

▪ What is least square criterion?

As the name implies, we are looking for this square deviation of Y
observation and then we are trying to fit a straight line in such a way
that the sum of squares deviation are minimum.
 Basic Information About Regression Process

❑ For the least-squares criterion, the magnitude of b can be calculated.

X
Basic Information About Regression Process

https://www.youtube.com/watch?v=nk2CQITm_eo
 Basic Information About Regression Process

❑ We want to ensure while fitting the straight line that the sum of the squared
deviation is minimum. If that is minimum, we can say that this straight line
represents more effectively all the points plotted on the graph. That is mean
by least square criteria square.

Y
𝐘𝐝

𝐘𝐝 𝒀𝒆 = a + bX

X
 Basic Information About Regression Process

❑ The parameter a can be estimated from the equation:

𝐘𝐞 = a + bX 𝑎 = 𝑌ത - b𝑋ത

Where,

= The mean of the X and Y observations, respectively.

𝑥 = X - 𝑋ത 𝑦 = Y - 𝑌ത
 Basic Information About Regression Process

❑ Using the summation symbol

https://www.youtube.com/watch?v=KbvD6FlIJGU
 Basic Information About Regression Process

❑ How to calculate linear regression using least square method?

https://www.youtube.com/watch?v=JvS2triCgOY
 Basic Information About Regression Process

𝑌4
Y 𝑌2 𝐲𝐝
𝐘𝐞𝟒
𝐲
𝐲𝐞
𝑌5 ഥ
𝐘

𝑌3
𝑌1

𝑋1 𝑋2 𝑋3 𝑋4 𝑋5

X
 Basic Information About Regression Process

❑ It can also be shown that:

σ 𝑦 2 = σ 𝑦𝑑 2 + σ 𝑦𝑒 2
Where,

σ 𝑦 2 = The total sum of squares of the deviations of the Y observations about the
mean value.

σ 𝑦𝑑 2 = Sum of squares of deviations of the Y observations from the regression

line.

σ 𝑦𝑒 2 = The sum of squares of the deviations of the estimated value of Y (𝑌𝑒 )

about the mean value (regression of sum squares).
 Basic Information About Regression Process

❑ The residual sum of squares, σ 𝒚𝒅 𝟐 provides a measure of variability of Y

observations about the regression value.

❑ If σ 𝒚𝒅 𝟐 is small, then the regression equation may be considered to fit the

observed data well.

❑ The ratio of the sum of squares explained by the regression to the total sum
of squares is known as the coefficient of determination, and is usually
denoted by symbol, 𝑹𝟐 .
 Basic Information About Regression Process

❑ The value of 𝑹𝟐 is given as:

σ 2
2
𝑦𝑒
𝑅 = 2
σ𝑦
❑ This is very important indicator of effectiveness of regression analysis.
Coefficient of determination will indicate the extent of accuracy involved
in regression analysis.
 Basic Information About Regression Process

𝑌4
σ 2
Y 𝑌2 𝐲𝐝
2
𝑦𝑒
𝑅 =
𝐘𝐞𝟒
𝐲 σ 𝑦2
𝐲𝐞
𝑌5 ഥ
𝐘

𝑌3
𝑌1

𝑋1 𝑋2 𝑋3 𝑋4 𝑋5

X
 Basic Information About Regression Process

Question: Considering the previous graph, what is the maximum and the
least possible value of R square?

Answer: If there is no deviation at all between the observed and regressed

values obviously, the denominator and numerator are going to be
same. So, the maximum possible value of R square is 1. On the
other hand, if the independent variable is not explaining the
dependent variable at all, the R square can be 0. So, the range of
values for R square is 1-0.
 Basic Information About Regression Process

❑ The ratio 𝑹𝟐 may assume a magnitude between 0 and 1. For a regression

for which 𝑹𝟐 =1, there would be no variation remaining that is
unexplained by the independent variable used in the regression equation.
That means, your independent variable is so effective and is able to
explain the whole of the variation of the dependent variable. It is very rare
to happen in practice, it is only an ideal situation, which can be understood
theoretically.

❑ When 𝑹𝟐 = 0, the independent variable used would not explain any of the
observed variation in the dependent variable, and there is no relationship
at all between your independent variable and the dependent variable.
 Basic Information About Regression Process

Question: In a particular case, what does , R square = 0.85, mean?

Answer: As we said that when R square value is 1, the independent

variable is fully explaining the variation of Y, when R squared is
0.85 obviously, the independent variable is able to explain
variation of Y to an extent of 85 percent; 85 percent of the
variation of the dependent variable is explained by the
independent variable chosen for regression analysis.
 Basic Information About Regression Process

❑ How to Calculate R Squared Using Regression Analysis?

https://www.youtube.com/watch?v=w2FKXOa0HGA
 Basic Information About Regression Process

❑ A second useful measure of variability of a regression line is the standard

error of the estimate, which is estimated as:

σ 𝑦𝑑 2
𝑺𝒆 =
(𝑛−𝑘)
▪ Where the denominator (n-k) is called the number of degrees of freedom
associated with the sum of the squares, σ 𝑦𝑑 2 .

❑ The number of degrees of freedom may be determined by subtraction, from

the number of observation, n, the number of statistics that had to be
calculated from the sample, to obtain the sum of squares.
 Basic Information About Regression Process

❑ Standard Error of the Estimate

https://www.youtube.com/watch?v=r-txC-dpI-E&list=PLF596A4043DBEAE9C&index=4
https://www.youtube.com/watch?v=A82brFpdr9g
 Basic Information About Regression Process

Question: In this particular case, how many statistics are to be calculated to

get sigma 𝒚𝒅 square?

Answer: Sigma 𝒚𝒅 is nothing but, deviation of the observed value of Y with

respect to the regression line. So, you need to get the regression
line, first to get the regression line you need to have at least two
statistics namely, 𝐗ഥ and 𝐘
ഥ, then only you are able to go further with
the values of small x and small y to calculate the regression
coefficients and then, get the equation for the regression line. That
is what is meant by number of statistics required to estimate the
value of sigma y d square.
 Basic Information About Regression Process

❑ For example, for a simple regression equation with one independent

variable, the two statistics to be calculated to obtain the sum of square are:

(𝑋ത ) and (𝑌)

ത
 Basic Information About Regression Process

❑ The regression coefficient b is a statistical estimate. Obviously, we estimate

the value of b, because there is no direct equation available to get accurate
value of b and it is, therefore, subjected to error and it is better to check for
the error associated with regression coefficient by some statistical measure.

▪ This standard error is a concept used to convey the magnitude of this

error, and is estimated from the following expression:

𝟐
𝑺𝒆
𝑺𝒆𝒃 =
Where, σ 𝒙𝟐
𝐒𝐞 = The standard error of estimate for the regression
σ 𝐱 𝟐 = The sum of squares of the deviation of x
observations about the mean.
 Basic Information About Regression Process

❑ In other words, it is nothing but, X minus ഥ X whole square, summed over all
the values. Now, the t test may be used to determine whether an estimated
regression coefficient is significant, by forming the following ratio:

෍ 𝑥2 = ෍ X − ഥ
X 2

t = The regression coefficient divided by standard error of the regression

coefficient.
 Basic Information About Regression Process

❑ The Normal Distribution and the 68-95-99.7 Rule

https://www.youtube.com/watch?v=mtbJbDwqWLE
 Basic Information About Regression Process

❑ Z-Scores, Standardization, and the Standard Normal Distribution

https://www.youtube.com/watch?v=2tuBREK_mgE
 Basic Information About Regression Process

❑ How to calculate t distributions

https://www.youtube.com/watch?v=QoV_TL0IDGA
 Basic Information About Regression Process

Example: Develop a trip production equation and calculate all the relevant
statistics to check the validity of the equation using the following data:

Average household size: 2 3 4 5 6

Average total trips made per day: 5 7 8 10 10

The value of t statistics for 3 degree of freedom at 5% level of significance is

“2.353” .

❑ As you know the second row values are dependent variables and first row
values are independent variables.
 Basic Information About Regression Process

❑ In this example the value of t statistics for 3 degree of freedom at 5% level

of significance is 2.353. This information is required, because finally to get
this statistical validity of any regression coefficient will be comparing the
table value of t with the actual calculated value. Here the table value given
as 2.353 at 5 percent level of significance, why 5 percent?, it can be
anything it depends upon the expected level of accuracy.

▪ Normally, in most statistical experiments, 5 percent level of significance

is acceptable that is how here the 5 percent comes in; you can aim for 1
percent level of significance, that means, 99 percent accuracy that is
also acceptable.
 Basic Information About Regression Process

𝐘𝐞 = a + bX 𝒚𝒆 = 𝐘𝐞 - 𝐘ത 𝑎 = 𝑌ത - b𝑋ത
σ 𝟐
𝟐
𝒚𝒆 𝐘𝐞 = a + b X
𝑹 =
σ 𝒚𝟐
ഥ )𝟐
𝒙𝟐 = (𝐗 − 𝑿

σ 𝑦𝑑 2 𝒚𝒅 = Y−𝐘𝐞
𝑺𝒆 = 𝑺𝒆 𝟐
(𝑛−𝑘) ഥ )2
𝒚𝟐 = (Y− 𝒀 𝑺𝒆𝒃 =
σ 𝒙𝟐
𝑺𝒆 < 𝑺𝒅 Hence , O.K.
σ 𝒚𝟐
𝑺𝒅 = 𝒃
𝒏 − 𝑲−𝟏 𝒕=
𝑺𝒆𝒃
 Practice!

ഥ X- 𝑿
Y- 𝒀 ഥ Y-𝐘𝐞 ഥ 𝟐
𝐘𝐞 - 𝐘
𝟐 𝟐 𝐘𝐞 𝒚𝒅 𝟐 𝒚
Y X (y) (x) 𝒚 𝒙 𝒙𝒚 (𝑦𝑑 ) (ye ) 𝒆

5 2

7 3

8 4

10 5

10 6

Σ 40 20

ഥ =?
𝒀 b =?
𝒀𝒆 = ?
ഥ=
𝑿 a=?
 Basic Information About Regression Process

ഥ X- 𝑿
Y- 𝒀 ഥ Y-𝐘𝐞 ഥ 𝟐
𝐘𝐞 - 𝐘
𝟐 𝟐 𝐘𝐞 𝒚𝒅 𝟐 𝒚
Y X (y) (x) 𝒚 𝒙 𝒙𝒚 (𝑦𝑑 ) (ye ) 𝒆

5 2 -3 -2 9 4 6 5.4 -0.4 0.16 -2.6 6.76

7 3 -1 -1 1 1 1 6.7 0.3 0.09 -1.3 1.69

8 4 0 0 0 0 0 8 0 0 0 0

10 5 2 1 4 1 2 9.3 0.7 0.49 1.3 1.69

10 6 2 2 4 4 4 10.6 -0.6 0.36 2.6 6.76

Σ 40 20 18 10 13 40 1.10 16.9
𝟒𝟎 σ 𝒙𝒚 𝟏𝟑
ഥ=
𝒀 =𝟖 b= = 𝟏𝟎 = 𝟏. 𝟑
𝟓 σ 𝒙𝟐
𝒀𝒆 = 𝟐. 𝟖 + 𝟏. 𝟑 𝑿
𝟐𝟎 ഥ −𝒃𝑿
ഥ
ഥ=
𝑿 =𝟒 a= 𝒀 a = 𝟖 − 𝟏. 𝟑 ∗ 𝟒 = 𝟐. 𝟖
𝟓
 Basic Information About Regression Process

𝟐 σ 𝒚𝒆 𝟐 𝟏𝟔.𝟗
𝑹 = σ 𝒚𝟐
= = 0.939
𝟏𝟖
❑ Now, let’s see how to go about using the value of Se and get some
inference and standard deviation is related to the observed value of Y :

σ 𝒚𝒅 𝟐 𝟏.𝟏
𝑺𝒆 = = = 0.6055
𝒏 −𝑲 𝟓−𝟐

σ 𝒚𝟐 𝟏𝟖
𝑺𝒅 = = = 𝟐. 𝟏𝟐
𝒏 − 𝑲−𝟏 𝟓 −𝟏

𝑺𝒆 < 𝑺𝒅 Hence , O.K.

 Basic Information About Regression Process

Question: What we really mean by 𝑺𝒆 and 𝑺𝒅 ? What is the standard

deviation in general?

Answer: Standard deviation is the level of scatter of observed values with

respect to the mean, and the standard error of estimate is nothing
but, again this scatter of the observations with respect to the
regression line instead of mean line.

❑ If the scatter of observed point with reference to a regression line is less

than the actual scatter of the observations with respect to its own mean, then
the regression line is relatively closer to all the points compared to the mean
of the observed points. So, if Se is less than 𝑺𝒅 we can say that, it is good.
𝑺𝒆 in this example is less than 𝑺𝒅 hence, it is O.K.
 Basic Information About Regression Process

❑ There is another way of looking at the correctness of your regression

analysis compare 𝑺𝒆 and 𝑺𝒅 value, which is to get the t value as regression
coefficient divided by the standard error of estimate of that particular
coefficient.

𝑺𝒆 𝟐 (𝟎.𝟔𝟎𝟓𝟓)𝟐
𝑺𝒆𝒃 = σ 𝒙𝟐
= = 0.1915
𝟏𝟎
𝑺𝒆𝒃 = Standard error of estimate of the regression coefficient

𝒃 𝟏.𝟑
𝒕=
𝑺𝒆𝒃
= 𝟎.𝟏𝟗𝟏𝟓 = 𝟔. 𝟕𝟖𝟗

❑ ‘t’ is greater than the table value of 2.353 @ 5% level for 3 degrees of
freedom.
 Basic Information About Regression Process

❑ In this example t is greater than the table value of 2.353 at 5 percent level of
significance for 3 degrees of freedom. According to t table the expected
value t as per the expectation of 5 percent of significance is only 2.353, but
our calculated value is much higher, so we can say that, the significance is
more than 5 percent.

▪ Now, the inference is the significance of b is much higher than the 5

percent level, the value is very high.
 Basic Information About Regression Process

❑ In this example we have regressed one independent variable against a

dependent variable.

❑ The independent variable was nothing but, the number of trips made for
any particular purpose or for all purposes, total number of trips made by
household for all purposes together, and The independent variable is just
the household size.
 Basic Information About Regression Process

Question: Would it be a practical situation? Do we really develop production

model relating trip made for all purposes and the household size?

Answer: It is not going to help us in practice. In practice, we will be

developing regression equations for trips made by households for
different purposes. It will have a set of e.g. five, six trip
production equations; one trip production equation for work trips,
another equation for educational trips, third one for shopping trips,
fourth for personal business trips and so on.

❑ As we said earlier, the set of independent variables that will influence these
categories of trips are going to be different or even if they are same, there
extent up influence of different trips type, that means the trip types will be
different.
 Basic Information About Regression Process

❑ In case of educational trips for example, trip production for the purpose of
education there likely independent variables are, number of students in the
household definitely that will be one of the variables then there could be
other causal variables like household vehicle ownership or income. Very
rarely, we just develop trip production equation with only one independent
variable. Normally, more than one independent variables are common in any
trip production equation.
❑ If you take trip production equation for shopping trips, then household
income might be an important variable plus could be level of vehicle
ownership. So, that is how you must identify and pick the causal variables to
develop trip production equation.
▪ So, the current example is not suited for real life situation, and this
example helped just to understand or demonstrate the analytical principle
involved in the regression analysis resulting in trip production equation,
this is meaningless as far as, practical application is concerned.
 Basic Information About Regression Process

❑ This is the example the average household size is 2, 3, 4, 5 and 6, they

cannot be independent household, because it is clearly stated this average,
average household size, average of how many households? How do we take
this average? These are zonal averages. We divide the entire urban area into
traffic zones, so for trip production analysis we deal with zonal average
values.
❑ So, here the average household size means, if there are 10,000 households
in a traffic zone then it is the average value of all the 10,000 observations
and if you look at the other variable, average total trips made per day for
that particular household the numbers given are whole numbers, but when
you work out these values in practice you may end up with fractions also.
You may end up with trips like 5.25, 7.101 and so on.
 Basic Information About Regression Process

❑ Because, you are taking averages, but you can work with these fractions
even though, they are not realistic and then later on finally, you can round
off the values.

❑ Since, these are averages you have to put up with an unrealistic numbers
and go ahead with your analysis and finally, round off the values to get the
appropriate numbers for subsequent analysis.
 Basic Information About Regression Process

❑ So far, we had single independent variable but in reality, you may have to
deal with more than one independent variable. So, what happens to the
regression equation, if you have for example two independent variables?

❑ Let’s see the equation for two independent variables case as, Ye to be equal
to a plus b1 X1 plus b2 X2, and as you see there are two independent
variables, X 1 and X 2 in this case.
 Basic Information About Regression Process

❑ Two Independent Variables

▪ The general form of regression equation for two variables case, can be
written as:
𝒀𝒆 = 𝒂 + 𝒃𝟏 𝑿𝟏 + 𝒃𝟐 𝑿𝟐

(σ 𝒙𝟐 𝟐 ) ( σ 𝒙𝟏 𝒚) − (σ 𝒙𝟏 𝒙𝟐 )(σ 𝒙𝟐 𝒚)
𝒃𝟏 =
σ 𝒙𝟏 𝟐 σ 𝒙𝟐 𝟐 − (σ 𝒙𝟏 𝒙𝟐 )𝟐
(σ 𝒙𝟏 𝟐 ) ( σ 𝒙𝟐 𝒚) − (σ 𝒙𝟏 𝒙𝟐 )(σ 𝒙𝟏 𝒚)
𝒃𝟐 =
σ 𝒙𝟏 𝟐 σ 𝒙𝟐 𝟐 − (σ 𝒙𝟏 𝒙𝟐 )𝟐

ഥ − 𝒃𝟏 𝑿𝟏 − 𝒃𝟐 𝑿𝟐
𝒂= 𝒀

https://www.youtube.com/watch?v=Ssp2tZKyiPY