Non-hierarchical grouping: `K-mean´ and `K-medoid´ of plaques cisterns in the Pajeu region - PE

Resumo

This work uses non-hierarchical grouping methods to evaluate the quality of the groups formed by plate cisterns according to some water quality variables. These methods use the cluster validation criterion to determine the optimal partition, which provides the most homogeneous groups possible. The methods were tested on a sample of 100 cisterns located in the Pajeú region. However, the non-hierarchical clustering method of `K-medoid´ formed more homogeneous groups, and thus the best performance according to the Silhouette [s (i) = 0.64] statistics.