SPATIAL CLUSTERING ANALYSIS TOOL DEVELOPMENT

• Main Author: LARASATI AYUDIANI (NIM : 13513025), VENNY
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/24621
• Institution: Institut Teknologi Bandung

Spatial clustering dealt with the grouping of spatial objects into clusters so that objects with high similarity are put into the same cluster and objects with low similarity are put into different clusters. There are two kinds of problems in spatial clustering, namely hotspot analysis and regionalization. For each of those problems, several approaches consisting of several different algorithms have been developed. Some analysis tools capable of conducting spatial clustering analysis have also been developed. However, those currently existing tools only implement a clustering technique for a certain problem using similar approaches. Therefore, different analysis tools need to be used in order to conduct spatial clustering analysis with different approaches. <br /> <br /> <br /> In this final project, a spatial clustering analysis tool that is able to act as a skeleton to various clustering approaches is developed. Existing statistics and algorithms from various approaches are analyzed. The design of the analysis tool is then made based on the analysis on the input, process, and output of those statistics and algorithms. The template method design pattern is utilized so that classes acting as a skeleton are designed as an abstract class. With this design pattern, the analysis tool will be easily extended in terms of new algorithms addition. <br /> <br /> <br /> Functional testing and nonfunctional testing is both conducted to evaluate the analysis tool. The functional testing is conducted using testing scenarios in order to check whether or not the analysis tool is able to function as it was defined. All testing scenarios have been conducted with correct results. The nonfunctional testing is conducted to test the extensibility of the analysis tool. This test is performed by implementing spatial clustering algorithms that were not included in the analysis. Those algorithms were successfully implemented thus the analysis tool is proven to be extensible.