Page 60 - Çevre Şehir İklim İngilizce - Sayı 4

Urban Transformation Forecasts with Artificial Intelligence
Based Algorithms

for the areas attributed as land which were sold in Gölbaşı district in 2017 and
the unit price estimate value of the land was defined as output variable. Within
the scope of the study, multiple regression analysis and ANN methods were
2
used. According to the results, coefficient of determination in ANN (r ) was
found to be more successful by exceeding the multiple regression analysis,
which was calculated as 94% to 89%.
A similar study was conducted by Güner (2021) for the province of Ankara. In
this study, ANN was used by taking into account economic data such as housing
prices, exchange rate, interest rates, consumer confidence index, industrial
production index, construction confidence index for 82 months between 2013
and 2019. According to the results, while the determination coefficient of
2
feedforward (r ) was 0.9955, this value was found to be 0.9863 in the Elman
neural network. According to the results, high accuracy was achieved in the
use of the feedforward neural network, and it was observed that it performed
well in predicting housing prices.
Wan and Shi (2021) studied on urban restoration public space design using
methods based on the convolutional neural network model. With this study,
the feasibility of using artificial intelligence learning, deep learning and other
related algorithms was investigated. The model, which used urban traffic
road network, urban neighbourhood spatial information and urban building
function data, is a case study in terms of creating appropriate urban areas and
an optimal road network for vacant spaces in cities.
Sipahioğlu and Çağdaş (2022) made urban growth modeling using scenario-
based cellular automation and ANN methods. In the model where land use
and driving factor data are used for simulation, six different development
scenarios were formulated according to the Strategic Plan of Izmir Province
and the effects of future urban planning strategies were identified.
As a result, it is seen that the ANN method is often used in the literature in
the planning of urban development and modeling of transformation studies.
However, in Türkiye, these studies have been carried out only in housing price
estimates and on a regional basis. Taking into account the economic data, the
modeling of urban transformation on a national scale and forecasts for the
future have been revealed with this study, and this deficiency in the literature
has been tried to be met with this study.

The Data, Methods and Prediction Model Studies Used

Within the scope of this study, dependent variables of the number of
apartments, number of licenses, surface area as well as the independent
variables of gross national product (GNP), inflation rate (%), interest rate (%)

Year 2 / Issue 4 / July 2023 49