APPLICATION OF PRINCIPAL COMPONENT ANALYSIS METHOD IN LAND PRICE MODELLING BASED ON COMMERCIAL FACILITY ACCESS
Geographically Weighted Regression method is a derivative method of Ordinary Least Squares Regression which can include local parameters in the form of the geographic location of the observation points in the estimation of the parameters of the regression equation, so that the equation formed at eac...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/72051 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Geographically Weighted Regression method is a derivative method of Ordinary Least Squares Regression which can include local parameters in the form of the geographic location of the observation points in the estimation of the parameters of the regression equation, so that the equation formed at each point varies with the geographic location. The GWR method can help modeling land prices to be more comprehensive because it can consider the characteristics of the surrounding land price and also the geographical conditions of the environment. However, involving many variables will cause the estimated parameters of the regression equation to become unstable because the dimensions are too high. Patterns and relationships between observations and between variables can be insignificant due to redundancy of predictor variables. Visualization and analysis of the predictor variables in the model will also be a challenge when there are lots of variables. The Principal Component Analysis (PCA) method can overcome this problem by reducing dimensions or simplifying the variables into fewer variables without removing important statistical information from the original set of variables. In this study, PCA was implemented to generate new variables that can be used for modeling land prices using the OLS and GWR methods. PCA produces three principal components (PC), each of which can summarize the pattern of variables at the study site. The three PCs were then used as predictor variables in the OLS and PCA models which resulted in RMSE of IDR 1,957,611/m2 and IDR 1,765,571/m2, respectively. Then a comparative analysis was carried out using the GWR method using the original predictor variables and other effects of the application of the PCA method on modeling land prices.
|
|---|