Predictive modelling in grape berry weight during maturation process: comparison of data mining, statistical and artificial intelligence techniques

Abstract

Environmental and geographical factors are two of the key aspects conditioning the growth of any crop, in such a way that the ability to predict significant variables of grape maturation can be highly useful to vine-growers. Berry weight is one of the variables monitored during this period, and the wineries have called for the development of an accurate prediction model. This study compares various types of data mining (DM) and artificial intelligence (AI) algorithms for developing an efficient prediction model for determining the variations in weight of grape berries during the ripening process according to the environmental and geographical properties not only throughout the ripening period but throughout the plant’s cycle. The final objective is the search for a model that is efficient for data for new years with different properties to those in the past. This model helps the grower to harvest the grapes on the most suitable date for producing the best possible wine.