CLUSTER ANALYSIS ON LAND CHANGE DATA THROUGH DIVISIVE HIERARCHICAL ALGORITHM CASE STUDY: EAST JAVA PROVINCE

• Main Author: Khotimatul Amanah, Luthfita
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/63865
• Institution: Institut Teknologi Bandung

Population growth has increased the need for changes in land use for various purposes that support the community's socio-economic development. This thesis observed changes in the land cover through cluster analysis in several districts/cities that are part of the province of East Java. Thus, it obtains various helpful information related to the causes of land cover changes, including population, distribution/distribution of rural-urban population, and other variables. The purpose of cluster analysis is grouping objects based on their similarity—the similarity from the size of the distance. The distance used is the Euclidean distance. Two clusters are said to be different or dissimilar if the two objects have a significant distance. The divisive hierarchy algorithm (DHA) in cluster analysis applies a top-down approach by assuming all members are in the same cluster. Then the cluster division is determined based on the dissimilarity of distance. Each type of land with similarity is naturally in a group, so it is the reason why AHD is applied to land cover changes in East Java Province by involving the physical condition of the land and its socio-economic status. There are 53,283,768 East Java Province land area pixels with 13 variabledriven factors. Four hundred samples were obtained using stratified random sampling. The results obtained in the land clustering process in East Java Province are 69.25% of land use changes with 21 land change patterns, and the remaining 30.35% are stable. The most significant change (21.75%) occurred in wetland agriculture to dry land agriculture spread over the western area of East Java Province. In addition, the evolution of land into a pond area also dominates in the eastern region of East Java Province. The conversion of function to built-up land is relatively small (1%). However, this is interesting because most of the built-up land are quite stable in the big cities of East Java Province.