Differentiation of mangoes (Magnifera indica L.) conventional and organically cultivated according to their mineral content by using support vector machines

Abstract

Mangoes of uniform genetics (Lippens variety) cultivated in the Gomera Island (Canary Islands) by conventional and organic farming were used to analyze the mineral content in order to differentiate crops cultivated in the same geographic area by the cultivation practices. Farming differences as well as soil differences may be reflected in the mineral content of the mangoes cultivated in these extensions. Concentration metal profiles consisting of the content of Ca, Co, Cu, Fe, K, Mg, Mn, Na, Ni and Zn in mangoes were obtained by using atomic absorption spectrometry (AAS). Pattern recognition classification procedures were applied for discriminating purposes. Linear discriminant analysis (LDA) allows to a classification performance of about 73% and support vector machines (SVM) found up to a 93% of prediction ability. The classification success when applying support vector machines techniques is due to their ability for modeling non-linear class boundaries.