ANALYSIS OF FACTORS AFFECTING LAND VALUE OF TRANSMISSION TOWER WITH GEOGRAPHICALLY WEIGHTED REGRESSION
The development of Electrical Infrastructure in the form of transmission lines for electricity requires land for the location of the tower site. The many locations of tower footprints are spread over a geographical area with different characteristics (spatial heterogeneity) which are thought to affect land values. Spatial heterogeneity is when the estimator parameters give different responses at different locations for the same explanatory variable. This study will determine the factors that affect the value of land on the tower footprint in Galang District using a geographically weighted regression model with a fixed gaussian kernel weighting. This study compares the observed land value with the predicted land value indication using the linear regression model (OLS) and the Geographicly Weighted Regression (GWR) model. This land value modeling uses 4 explanatory variables, namely land distance to the nearest road (X1), land distance from the CBD (X2), land distance to main transportation routes (X3) and land elevation (X4). The results showed that the minimum observation land value was IDR. 78.000 m2 (2021), the minimum GWR model predicted land value is IDR. 85.000/m2 and the minimum predictive land value of the OLS model is IDR. 86.000/m2. While the maximum observation land value is IDR. 130,000 m2, the maximum predicted land value of the GWR model is IDR. 125.000/m2 and the maximum predicted land value of the OLS model is IDR. 123,000/m2. In general, the difference between the GWR model land value and the observed land value is IDR. 5.562/m2 is smaller than the difference in the value of the OLS model land to the observed land value of IDR. 25,020 m2. So in general the GWR prediction soil model is closer to the regression line of the observed land value compared to the OLS model prediction value line. GWR model with fixed kernel weighting function gaussian has an R-square value of 75.98%, AIC -104.63 and a land value difference of IDR. 5.562/m2 with an observation value while the OLS model has an R-square value of 73.74% and AIC -102.80 and the difference in land value is IDR. 25,020/m2 with observation value. So that the GWR model is better than the OLS model in modeling the land value of the tower footprint in Galang