Resumo

NAVA-NAVA, Adan  e  ANTUNEZ, Pablo. Application of quantile regression to predict stem volume: Case study. Ecosistemas y recur. agropecuarios [online]. 2018, vol.5, n.15, pp.591-600. ISSN 2007-901X.  https://doi.org/10.19136/era.a5n15.1498.

In estimating the stem volume of trees, it is desirable to have precise mathematical tools in which the least amount of inputs is incorporated to achieve adequate control over the extraction of roundwood from a forest under exploitation. The objective was to determine whether quantile regression can more accurately predict the stem volume of Pinus patula Schiede ex Schltdl. & Cham than conventional techniques, using diameter at breast height as a predictor. Data from 148 dominant and co-dominant trees were used to generate three quantile regression lines. The fitting and application of the proposed regression were compared with that obtained with the Berkhout exponential model with fitting by non-linear least squares. The results suggest that this technique improves the volumetric prediction of the trees by generating lines in different segments of the conditional distribution, allowing a significant reduction in the prediction error for the data far from the arithmetic mean.

Palavras-chave : Volume equations; heterogeneous data; input reduction; homoscedasticity of residuals.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )