Alternative methodology for the cluster analysis

The objective of this work was to propose an alteration in the clustering procedure accomplished by the Tocher method, changing it from a simultaneous to a sequential method. Accession collections were generated through simulation to illustrate and compare the original and the sequential methodologies. In each collection, different characteristics for individual data as well as experiments with repetition were evaluated. The simulations and the statistical analyses were accomplished using the GENES applicative. The number of groups formed with the sequential method was smaller than the number formed with the original Tocher method. The genotypes clustering were not influenced by the most distant genotype when the sequential method was used. The increment limit in the mean of the intra-group distance, estimated after the formation of a new group, is an estimative of dissimilarity between the group's accessions. The sequential Tocher method grouped genotypes comprising higher dissimilarity with more effectiveness than the original Tocher method.