Cocoa crop forecast based on pod counting method

Abstract

The development of the method is based on the relationship between cocoa crop production and pod observation. A study of the pod profiles during a crop year is used to create the predictive point when the pod profiles are classified in homogeneous clusters and to verify the predictive model relevance when the prediction is an interpolation or an extrapolation. Predictions with lower estimation errors are obtained three to six months before each crop end.