| Summary: | 碩士 === 國立成功大學 === 都市計劃學系碩博士班 === 91 === The construct of the land price depends on how people use it. People use many kinds of method and model in order to forecast the exact land price. There are two kinds of land price assessment: 1.The assessment that involves the private rights 2.In order to levy the tax or compensation for private person that hold by the government. The process of these two kinds of assessment is very different because of the different aims of the assessments. We usually treated factors that effected the land price as a stable trend, that is, we ignore the spatial character of the factors that effect the land price. The latest researches shows that spatial non-stationarity is an important point of view for land price assessment, for that reason, we developed the land price assessment model to reflect the spatial variation.
Spatial statistics is one branch of the statistics system that stress the spatial variation, this thesis use “Geographically Weighted Regression” which is the latest method of spatial statistics and we can avoid the disadvantages of treat factors as a stable spatial variation drift. The hardest factor to treat is that “location factor”, we usually delimit the land into some zone that has the same land price character before, but it is hard to choose that what factor should be excluded. This thesis uses Kriging method that categorized in Geostatistics system to discuss this issue.
Kriging method is a popular method in Geostatistics system. We can find the “Spatial Variation Angle” and “Range” of the land price in space to prove the spatial non-stationarity in Tainan city east district, middle district and west district. We also use the “Range” in Kriging method to choose the location factors and the spatial bandwidth in Goegraphically Weighted Regression.
The aim of the thesis combined these two methods is to reflect the spatial non-stationarity that how the factors effect the land price and overcome the disadvantages of stable land price assessment model. The new land price assessment model shows a good result in R2 and land price assessment efficiency.
|
|---|
