New method for representing multidimensional social phenomena: smart k-means

This exploratory research proposes a new method for representation of multidimensional phenomena based on the k-means algorithm. The so-called "smart k-means" uses the Shannon entropy index to identify and discard individual indicators with low informational diversity, which do not distinguish the decision-making units adequately. This disposal process ensures groups with acceptable cohesion and resolution, and within the number of group arrangements desired by the decision-maker. The results of the application of the new method in the representation of social exclusion in Maringá show that the process of discarding individual indicators of low informational diversity ensures the achievement of internally homogeneous and externally heterogeneous groups. These results contribute to a reliable representation of the multidimensional phenomenon, facilitating its understanding and the development of public policies to reduce it.