Learning Preferences in a Spatial Multiple Criteria Decision Approach: An Application in Public Security Planning
Ciro José Jardim de Figueiredo, Caroline Maria de Miranda Mota · International Journal of Information Technology & Decision Making · 2019
This study presents a model to identify and classify vulnerable places regarding violence in public areas. The model considers multiple objectives and multiple viewpoints by using a graphical visualization for exploring vulnerability. The methodology is supported by a Dominance-based rough set approach in conjunction with preference learning and Geographic Information Systems, and requires the use of decision-makers’ (DMs’) previous knowledge for holistic assessment to get individual results. We considered an original approach for aggregating those individual results to obtain a recommendation from the final output. The preferences are assessed interactively to decrease the cognitive effort of each DM, starting from a small subset of holistic evaluations and expanding the process by incrementing new information in each stage of the model. We also assigned indicators to identify the quality of the results, participation in the individual preferences, and to avoid inconsistencies. We used the model to identify areas that merit more resources to combat crime and employed several criteria that assess types and levels of violence. These results may help policy-makers and planners to draw up and refine public policy interventions in the public security context.