[-] Show simple item record

dc.contributor.advisorHaithcoat, Timothyeng
dc.contributor.authorTanrikulu, Ferideeng
dc.date.issued2023eng
dc.date.submitted2023 Springeng
dc.description.abstractValuation predictions are used by buyers, sellers, regulators, and authorities to assess the fairness of the value being asked. Urbanization demands a modern and efficient land valuation system since the conventional approach is costly, slow, and relatively subjective towards locational factors. This necessitates the development of alternative methods that are faster, user-friendly, and digitally based. These approaches should use geographic information systems and strong analytical tools to produce reliable and accurate valuations. Location information in the form of spatial data is crucial because the price can vary significantly based on the neighborhood and context of where the parcel is located. In this thesis, a model has been proposed that combines machine learning and spatial context. It integrates raster information derived from remote sensing as well as vector information from geospatial analytics to predict land values, in the City of Springfield. These are used to investigate whether a joint model can improve the value estimation. The study also identifies the factors that are most influential in driving these models. A geodatabase was created by calculating proximity and accessibility to key locations as well as integrating socio-economic variables, and by adding statistics related to green space density and vegetation index utilizing Sentinel-2 -satellite data. The model has been trained using Greene County government data as truth appraisal land values through supervised machine learning models and the impact of each data type on price prediction was explored. Two types of modeling were conducted. Initially, only spatial context data were used to assess their predictive capability. Subsequently, socio-economic variables were added to the dataset to compare the performance of the models. The results showed that there was a slight difference in performance between the random forest and gradient boosting algorithm as well as using distance measures data derived from GIS and adding socioeconomic variables to them. Furthermore, spatial autocorrelation analysis was conducted to investigate how the distribution of similar attributes related to the location of the land affects its value. This analysis also aimed to identify the disparities that exist in terms of socio-economic structure and to measure their magnitude.eng
dc.description.bibrefIncludes bibliographical references.eng
dc.format.extentx, 62 pages : illustrations (color)eng
dc.identifier.urihttps://hdl.handle.net/10355/96184
dc.identifier.urihttps://doi.org/10.32469/10355/96184eng
dc.languageEnglisheng
dc.publisherUniversity of Missouri--Columbiaeng
dc.relation.ispartofcommunityUniversity of Missouri--Columbia. Graduate School. Theses and Dissertationseng
dc.titleLand valuation using an innovative model combining machine learning and spatial contexteng
dc.typeThesiseng
thesis.degree.disciplineData Science and Analytics (MU)eng
thesis.degree.grantorUniversity of Missouri--Columbiaeng
thesis.degree.levelMasterseng
thesis.degree.nameM.S.eng


Files in this item

[PDF]

This item appears in the following Collection(s)

[-] Show simple item record