Attesting compliance of biodiesel quality using composition data and classification methods

Four methods for direct classification (decision tree classifier, K-nearest neighbors, support vector machine and artificial neural networks) have been optimized and compared to classify biodiesel samples according to their compliance to viscosity, density, oxidative stability and iodine value, having as input the composition of fatty acid methyl esters. The eight databases for the classification of each property according to the standards EN 14214, RANP 45/2014 and ASTM D6751 were increased by synthetic minority over-sampling technique to deal with imbalanced data (different numbers of compliant and non-compliant samples). Taking as criteria the best performance and lowest complexity, decision tree classifier and K-nearest neighbors methods presented the best results, ensuring maximum accuracy, sensitivity and specificity (1.0000) for samples of external validation, for almost all properties and standards. A comparison between these methods of direct classification and empirical equations (indirect classification) distinguished positively the direct classification methods in the problem addressed.