Abstract
In order to predict sugar beet yield reduction from weeds, obtained data was fitted to some non-linear regression functions. Two independent variables were selected for validating these functions: time and thermal time. The hypotheses of normality, homoscedasticity, independence of residuals, predictive capacity or goodness of fit with respect of the model, and diagnosis of the model, have been verified by means of residuals analysis for both independent variables. In early competition, a hyperbolic function was rejected due to the fact that some parameters seemed insignificant. However, the goodness of fit was similar to other functions. In late competition, some parameters were insignificant, but this hyperbolic function reaches the best goodness of fit. Comparing the validated functions, the logistical equation that includes the inflection point with independent variable time is the one that reaches the best results. The high flexibility of the model may allow the detection of special cases, and thus minimize risk. This study could be done anywhere in the world, as long as the mathematical model is adapted to the climate study area.
Keywords:
sugar beet; yield description; competition models; model discrimination
Thumbnail
