Page 32 - NGTU_paper_withoutVideo

Modern Geomatics Technologies and Applications

Fig. 4. Anselin Local Moran’s I analysis of COVID-19 Incidence rate across Iran, by the end of October 21, 2020

3.5 OLS regression model evaluation
Due to the rules announced by OLS model, explanatory variables which were highly correlated, removed from the
model. Variance Inflation Factor was used to determine if there is multi-collinearity between variables. Finally, among 10
potential explanatory variables, the most uncorrelated variables were considered as the selected explanatory variables for
inserting into the model. These two variables were percentage of the urban population and the percentage of 60 years and over.
Multi-collinearity was assessed through the variance inflation factor (VIF), which measures redundancy among explanatory
variables. Explanatory variables associated with VIF values larger than about 7.5 indicated that these variables are providing
similar information, and they were removed one at a time from the model based on VIF value until the model became
unbiased. As can be seen in Table 2, the result of OLS method indicates low multi-collinearity between selected explanatory
variables, due to the value of VIF (1.11), which is lower than 1.5.
2
The Adjusted R of OLS global model, was 0.55, which shows that almost 0.45 of the COVID-19 incidence rates across Iran
are happened by unknown variables to the model. These variations could be local, which not considered by the Global OLS
model. Besides, the value of corrected AIC for OLS, was 44.273. Among the models, the best model is the one with maximum
2
Adjusted R and minimum AICc. Coefficient, which describes the strength and type of relationship between each explanatory
variable and the dependent variable, was 0.07 and 0.47 for the percentage of the urban population and the percentage of
population aged 60 and over, respectively.

Table 2 Summary statistics of the OLS model on selected explanatory variables

Variable Coefficient St. Error T-statistic Probability VIF
Intercept -0.254792 0.15328 -1.66219 0.109032 ---
Urban population 0. 07694 0.00188 4.58922 0.00001 1.118491
Population 60+ 0.47955 0.01495 3.20600 0.00007 1.118491

7