Page 720 - NGTU_paper_withoutVideo

Modern Geomatics Technologies and Applications

As A is the transition matrix of the region, is the sum of the regions from i to j over the years from the starting point to the

target simulation courses and n is the number of land use types[20].
3.4 Cramer’s V

The Cramer's V method is a statistical method that shows that although the potential explanatory value of a variable is
good, it does not guarantee strong performance because it cannot guarantee the mathematical needs of the modeling approach
used and its complexity. If the value of Cramer's V is very low, the variable can be omitted. For good modeling, there should not
be much dependence between the variables.

2
2
The test can also be used to show the dependence between two variables, but the test is very sensitive to the size
of the sample [21]. For this reason, Cramer's V test is used to measure the strength of the relationship between variables [22].
2
The V-Cramer test can be taken directly from the statistical test.
In this method, the number between 0.1 to 0.4 represents the factor with the desired effect, higher than 0.4 indicates
the factor with the very desired effect and lower than 0.4 indicates the factor with the adverse effect or no effect.

3.4 Comparison matrix
Usually, two criteria of overall accuracy and Kappa index have been used in many studies to express simulation
accuracy, which indicate the correspondence between reality and simulated model. One of the advantages of the Kappa index is
the use of all comparison matrix values to calculate accuracy [16]. How to calculate the overall accuracy and Kappa index is
shown using the comparison matrix in Equations (8) and (9):
∑ =1
= (8)

∑ =1 − ∑ =1

= (9)
1 − ∑
=1
In the above equations = 1.2 … . represents the land cover classes in the region and represents the pixels of the

land cover classes in reality, which in the simulated model are also in the land cover classes . is the total number of

ℎ
ℎ
pixels and represents the pixels of the land cover classes are actually in the simulated model in the land cover classes,

ℎ
ℎ
and represents the sum of the class pixels in reality and represents the sum of the class pixels in the simulated
ℎ

ℎ

model.
4. Experimental results
4.1. Editing land use map
The existing maps for 1996 and 2006 had 21 classes and the 2016 map had 18 classes that needed to be re-classified,
and finally, the maps were produced with the same classification. This re-classification was done in the ArcGIS environment.

4