Feature extraction and classification of Chilean wines

In this work, results of Chilean wine classification by means of feature extraction and Bayesian and neural network classification are presented. The classification is made based on the information contained in phenolic compound chromatograms obtained from an HPLC-DAD. The objective of this study is to classify different Cabernet Sauvignon, Merlot and Carménère samples from different years, valleys and vineyards of Chile. Different feature extraction techniques including the discrete Fourier transform, the Wavelet transform, the class profiles and the Fisher transformation are analyzed together with several classification methods such as quadratic discriminant analysis, linear discriminant analysis, K-nearest neighbors and probabilistic neural networks. In order to compare the results, cross validation and re-sampling techniques were used.