Multi-level Data Fusion Strategies for Modeling Three-way Electrophoresis Capillary and Fluorescence Arrays Enhancing Classification of Wines

MARÍA S. DI NEZIO, MARCELO F. PISTONESI, WALLACE D. FRAGOSO, MÁRCIO J.C. PONTES, HC. GOICOECHEA, MÁRIO C.U. ARAUJO E BEATRIZ S.F. BAND.; AZCARATE S; CAMIÑA J; GOICOECHEA H C

Workshop; XI Workshop de Quimiometria; 2020

Over the years, analytical methods and dataanalysis tools commonly used in food qualityand process control had to be re-evaluatedand modified to fit these new tasks. In thisprogression of gathering more and betterinformation, the multivariate statistical analysisof fused data has become a powerful tool forenhancing the reliability of the results1. Beingthe key point how the information sources canbe combined to provide the joint classificationprediction of the samples, three levels of datafusion (DF) have been reported2. The aim ofthis work was to develop multiple strategies toassess the three DF levels on two secondorder arrays, with different data complexity, inorder to know the correlation and analogybetween both information sources for twofoldclassification purposes. Thus, the challengeconsisted in finding the optimal combination ofdata preprocessing, fused data and datamodeling that would provide the best results