Detection of Buildings in the Urban Transformation Process with Spatial Data Mining Methods

Abstract

Since the 1950s, with the acceleration of industrialization in Turkey, a severe demand for labor has occurred, especially in big cities, and therefore a rapid internal migration movement has emerged. As a result of the migration movement, irregular growth and settlement activities started in big cities. As a result of this distinctive settlement, many problems have emerged, especially infrastructure problems. Urban transformation projects have an important place in the solution of these problems. Urban transformation projects are a process that starts with the announcement of the area and ends with the transfer of the title of the citizen. According to the location of the cadastral parcel that the city owns, the closest building to the parcel is drawn to determine which flat to live in. Selecting the nearest building to it is done by human hands. This situation negatively affects the process in terms of both speed and accuracy. Spatial data mining-based clustering is significant for big data because it can automatically classify data. Within the scope of this study, the building identification process with the human factor has been automated using data mining-based spatial clustering methods, K-Means, DBSCAN, and OPTICS algorithms. As a result of the experimental evaluations, it was determined that the OPTICS clustering algorithm gave the most accurate result.