The objective of this work was to fit equations for total biomass of Guadua bamboo plants and to compare adjustment of equations by linear regression with data mining technique. Thirty-eight culms of bamboo were used, in which the following variables were measured: diameter at breast height, stem diameter, stem and its height. The stems were felled and weighed by using the destructive method. The actual biomass of 25 stems was used to fit the equations by the least squares method and 13 stems were used for the validation of the best equation. The fractions of biomass per compartment differ significantly (p < 0.05) among each other. The largest fraction of the biomass corresponds to the stem, representing 69.2% of the total, followed by rhizomes, leaves and branches, with 15.7, 10.8 and 4.2%, respectively. The best fitted equation for total biomass showed a correlation coefficient of 0.93 and standard error of estimate of 15%. On the other hand, the data mining technique showed a correlation coefficient of 0.81, with a standard error of 23.8%. Biomass of Guadua bamboo can be accurately estimated either by linear regression or data mining. In this work, the regression method performed better. Data limitations can be the key factor for the poor performance of data mining technique since once it requires a larger data set to function satisfactorily.
biomass; correlation; data mining; linear regression; Poaceae
