Knowledge-Based Multicriteria Spatial Decision Support System (MC-SDSS) for Trends Assessment of Settlements Suitability

Abstract

Spatial data mining is the discovery of interesting hidden patterns and characteristics that may be implicitly in spatial databases. This paper aims to produce a descriptive model for examining the suitability in settlements by applying various machine learning techniques to figure out the knowledge discovery in spatial databases (KDSD). The study illustrates the unique hallmark that characterizes the spatial data mining by conducting the data mining algorithms. Moreover, the study presents the importance of spatial data mining and discussed multiple data sets preprocessing, classification functions, clustering and outlier detection in directions supervised learning for extracting classification rules and assessing the local amenity based on rules reliability. The classification accuracy among the three methods of the classifier algorithms (Decision Tree, Rule-Based, and Bayesian) is also compared, thereby determining the most suitable classifier by experiments performance evaluation of the training and test set.