Thesis Type: Postgraduate
Institution Of The Thesis: Istanbul Technical University, Bilişim Enstitüsü, Bilişim Uygulamaları, Turkey
Approval Date: 2020
Thesis Language: Turkish
Student: BİLGE YILMAZ
Consultant: Uğur Algancı
Abstract:
As it well known,
world’s population keep rises day by day and the citys’ land- use change
becomes inevitable becuese of this population growth. This situation makes
difficult to carry out planning studies systematically in cities which aims to
establish healthy sustainable relations. It is nearly impossible to have
estimations on travel demand and attraction that consist of socio-economic
factors which have variable values and are affected by the social,
environmental and economic activities in the city. Today’s studies stand on the
methods that allows to make the most accurate estimation and these travel
demand forecasting methods generally require data collection and data counting
on the city scale or on the specific point in the city. The variability of the
data set causes repeating the estimation steps repeatedly and this demands
time, workforce and high budgets. In addition to the variable data set, there
is no comprehensive study on the extent to which socio-economic factors
associated with travel demands affect travel production and attraction, so it
is not possible to estimate the extent to which changes in travel demands
arising from an increase or decrease in these values. Because of these reasons,
it needs to be developed new travel demand forecasting methods which will
demand data and labor requirements less and can be easily updated.
Additionally, it needs to be determined the significance levels of the travel
demands for each point of the city, the parameters which effects these demands,
and to create analysis maps of these travel demands. With the transportation
planning activities, it is aimed to develop sustainable cities by evaluating
the studies on the prediction of travel demand. Thanks to these methods,
neighborhoods and avenues which have the most dense pedestrian and traffic
volume, and many infrastructure facilities which are needed will be able to
detected easiliy, therefore the cities will develop in acontrolled way.
The analytic hierarchy
process, which is one the multi-criteria decision making methods, which is an
increasingly popular application method, is used in studies which contain more
than one parameter, and the significance levels of these parameters is
determined by the intuitive or logical views of experts or non-experts in this
field and the synthesis of these collected data. The inverse distance weighted
method which is a deterministic procedure, and the ordinary kriging
interpolation methods, which are adapted to the linear and gaussian theoretical
variogram model, and also these are geostatistical procedure, make weight-
dependent weighting based on the principle that the objects close to the each
other will be more similar than to the objects are distant. It makes
predictions abput unknown points by using the observation points whose values
is known,and generates prediction maps for the entire surface of an area
interpolated throgh geographic information systems. With these methods, which perform
estimates with a lager number of sampling points, susccessful results were
obtained in applications for different disciplines such as river water quality,
regional landslide hazard, spatial precipitation distribution, estimates of
soil variables and such optimization problems or reserve estimation. Regression
analysis isan applications in which the relationships between dependent and
independent variables are explained by a mathematical fuction with numeratical
values, and the cross correlation technique is the application where the error
rates of the criteria whose value estimates are made are determined. Two of them
is frequently used in the verification steps.
Within the scope of
this thesis, which aims to contribute to transpostation planning studies and to
be open to dispute and advance, Istanbul has been chosen as a case study area
because it is a metropolis where an changes occurs constantly.In the study,
data sets consist of socio-economic factors related to Istanbul were classifies
according to standard deviation values and analysis maps were produced by using
geographical information systems. Thanks to broad literature review about on
the travel demand and attraction estimation methods which has been studied, it
is determined that the most used parameters that effects travel demand values
are numbers of population, average household income, vehicles, schools, employees
and students. To determine the significance level of these parameters, two
hundred particitpants from different participator groups were brought together
and their opinions gathered in logical manner through the analytic hierarchy
process.It was determined that the significance level of these parameters on
the age range is 25-59, 0-24, 60-75, 76-90+ respectively, and then other
characteristics comes respectively, number of employees, and students, job
opportunity, income status, population, number of vehicles and number of
schools. Classifications that are made by the standard deviation values in the
analysis maps and significance level of the parameters was used on the analysis
of the trip demand model value table. Analysis on both the model and the travel
demand values on 2012 were compared with the spatial interpolation methods.Similar
images were obtained from the analyzes made with two different data sets, the
disadvantages and the advantages of the methods in the analysis maps were also
explained.In the verification step, 20 districts were selected among 39
districts of Istanbul,by using the cross correlation method. The values of 20
points are removed from the analysis maps.With the travel demand datas of 38
districts, spatial interpolation methods were analyzed for 20 observation
points. The error ratio of the estimated values were calculated according to
the actual values. In that way, it has been detected which method makes
estimates with the least and most error rate for any district.Finally,
regression analyzes were carried out to be able to explain the correlation
between the predictive values and actual values of 20 observation points. It is
observed that for which districts of
Istanbul can be produced reliable values.
The data set used in
this study were prepared according to datas of 2012, which are transportation
network and property- line maps of the districts of Istanbul, trip demand and
attraction values, average household income, number of vehicles, employees,
employment volume, number of students, schools and 0-24, 25-59, 60-75, 76-90+
age groups of population.
As a result of this
study, thanks to the opinions received from 200 people, it was determined that
the weight of the travel demand factors determined by AHP approach directed the
model correctly. It has been identified that the travel demand model which is
obteained reflects the current station in 2012. And also, it has been
identified that the value estimates made by interpolation methods have a very
low error rate for some districts of Istanbul.So it is likely to say that the
method can be used, and developed by considering future suggestions to move
this study forward.