THE MODELING OF LAND PRICE PREDICTION WITH ADAPTIVE INDIVIDUAL GEOGRAPHICALLY WEIGHTED REGRESSION ALGORITHM AND COMPOUND INTEREST MODEL (CASE STUDY: BANDUNG TIMUR)

• Main Author: Puspa Handayani, Alfita
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/66636
• Institution: Institut Teknologi Bandung

Future land pricing information is needed to provide measurable information for strategic planning decisions and other long-term needs related to funding, extending credit and investment. This research was conducted to obtain information on future land prices through land prices prediction model. Prediction model is carried out using Geographically Weighted Regression (GWR) approach, which is able to describe the heterogeneity that occurs in an area and Compound interest model (CIM) approach which is a model for predicting future land prices based on the interest level. GWR is developed by applying an adaptive individual bandwidth algorithm with Inverse distance weighting (IDW) interpolation so that each point has an optimum bandwidth which results in a smallest error (GWRi). CIM is used with macroeconomic factors, namely interest rates that are predicted using a moving average. The two models are then integrated (GWRi-CIM) so that future land price predictions can be obtained. From this research, it was found that 11 spatial variabels have significant effect on land prices in East Bandung. These eleven variabels are variabels of security facility, health facility, commerce facility, basic education facility, higher education facility, terminals, toll gates, roads, topography, correctional facility, and slopes. Within study periods of 2007-2020, the phenomenon of economic change is more dominant than the phenomenon of spatial change in driving land prices in East Bandung. The novelty of this research lies in the GWRi, which is the development of an adaptive bandwidth algorithm to determine the optimum bandwidth at each point. An adaptive individual bandwidth algorithm (GWRi), increasing the ability of GWR from an error amount of Rp129,944/m2 to Rp612/m2. Another novelty is the integration between the GWRi and CIM models. The integration is carried out using the factor of future price increases due to spatial changes in the GWRi model and future price increases due to economic conditions from the CIM model. The integration of GWRi and CIM (GWRi-CIM) models increases the accuracy rate by 70% better than models using GWRi and 10% better than models using CIM.