Learning curves for the modeling of sugar cane productivity

Abstract

Predicting the final yield of a crop is one of the most important aspects of a mill's agricultural planning. However, numerous factors must be considered to ensure a realistic forecast. Data mining techniques are capable of generating models that predict these values by relating a large amount of data. In this project, we studied learning curves, a tool used in the analysis of a model's performance according to the amount of data available. In an analysis of a database for a sugarcane production, we compared three different modeling techniques, suitable for regression models in the prediction of the final productivity.