Hazards affect people regardless of their socioeconomic situation. This study proposes a computational method for data analysis in terms of the number of social vulnerability variables and samples of the case study in the Merapi proximal villages. For this purpose, the Self Organizing Map (SOM) is considered as an effective platform to identify the sites according to their similarity and to determine the most relevant variables to characterize the social vulnerability in each cluster; while Social Vulnerability Index (SoVI) is used for vulnerability index creation to measure the level of vulnerability. The dataset used for this analysis consisted of 12 variables, which represent the socioeconomic concepts, and collectively represent the situation of the study area, based on fieldwork conducted on September 2013. However, some of the variables employed in this study might be more or less redundant. The results showed that quantification and topographic errors presenting degree of accuracy of the representative data samples arranged in hexagonal map units varied considerably depending on the map size of the SOM. This indicated that some data samples require the removal of redundant variables. When we investigated the relative importance of variables in the reduced dataset, the parameters related to the number of migrate-in population (MOVEPPLIN) and the number of females (PRCTFEM) had the most significant impacts on the social vulnerability. From this study, we demonstrated that the SOM approach provided reliable estimates of clustering and the most significant variables, while SoVI works well in ensuring that positive value indicates high vulnerability, and vice versa.
Maharani, Y. N., Lee, S., & Ki, S. J. (2016). Social vulnerability at a local level around the Merapi volcano. International Journal of Disaster Risk Reduction, 20, 63–77. https://doi.org/10.1016/j.ijdrr.2016.10.012