Page 59 - Çevre Şehir İklim İngilizce - Sayı 4

Hacı Abdullah Uçan - Tayfun Dede - Sinan Nacar

Urban growth simulation was performed Guan et al. by (2015) with the
integration of ANN, cellular automation, restricted cellular automation. In
this study, geographical and spatial land use data (elevation, slope, distance
to the city center, railways, highways, national highways, number of cellular
urban areas) were used for Bejing in China in the periods between 2001-2005,
2006-2010 and 2011-2015. Urban growth prediction was simulated in three
socioeconomic-based forecast scenario studies.
The estimation of the growth of urban areas for the Greek city of Athens
was made by Triantakonstantis and Stathakis (2015) using the ANN. The land
change data (CORINE) between 1990-2000, altitude, slope, distance to the road,
distance to urban areas data were used as the inputs of the model. Cramer’s
V value was used for evaluating the land use classes for urban land use maps
of the year 2000. It was used in the verification phase for the calculation of the
urban forecast map of 2006 through the Markov Chain prediction process.
Housing prices for Eskişehir province were estimated by Yilmazel et al.
(2018), through ANN by taking into account factors such as the area of housing,
the number of floors, heating system and usage type. A machine learning
algorithm was created with the WEKA 3.8.0 program, the backpropagation
algorithm was used as the training algorithm, and the sigmoid function was
used in the hidden and result layers. 19 models were created according to
the number of 2-20 neurons and the correlation coefficient (R) of 0.9219, the
root mean squared errors (RMSE) of 0.192 and the mean absolute error (MAE)
values of 0.1441 were found with the highest Model 4. It is stated that the
study has the largest sample size of 5556 among the studies conducted in
Türkiye on a provincial basis.
Purevtseren et al. (2019) made forecasts for future zoning projects for the
year 2046 with the use of ANN based on the scenario of urban development
for the city of Erdenet, Mongolia. Remote sensing-based land use data and
satellite maps were used as inputs of the model. In the study where multi-layer
ANN was used, there is one entrance, one exit layers together with multiple
hidden layers. An accuracy rate of up to 81.2% was achieved at the 10,000
repetition level. Spatial forecasts were determined for Erdenet city for the years
2026, 2036 and 2046 according to the instantaneous developing scenario,
environmentally protected scenario and resource protection-based scenarios.
A pilot study was conducted by Ilhan and Öz (2020) in Gölbaşı district of
Ankara Province through a collective evaluation of real estate by the ANN
method. 13 parameters (location, KAKS, plan function, parcel shape, parcel
area, view, infrastructure, being on the Special Environmental Protection border,
ownership status (full / share), slope, road front, being a corner parcel, relevant
legislation under which the sale was made), were determined as input variables

48 The Journal of Environment, Urban and Climate