The use of models to describe the growths of plants has a wide application in the agronomy and cattle research. In this type of research, two problems are often present: to find the model that shows more information, which is best fitted to the biologic phenomenon and identifying curves that are statistically equal. Thus, the objective of the present work was to compare the quality adjustments models of non linear regression in the description of growth and to compare the curves from this best model in relation to different treatments. Data from an experiment installed in a factorial arrangement 2x5, in a randomized design of blocks with eight repetitions were used. The levels of treatments were constituted by two banana tree cultivars and five doses of a growth regulator. The cluster analysis was used with three statistics that allowed access to the ideal number of groups. In the comparison of growth models, the superiority of the logistic model was observed. For comparison of treatments, it was verified that the cluster method analysis with the changes proposed allowed to detect three groups of treatments, that are similar to the biologic behavior.
Musa spp.; growth analysis; non-linear models; cluster analysis; logistic model
