commercial instant coffee classification using an electronic nose in tandem with the comdim-lda approach

Coffee is an important commodity for brazil and ensuring product quality is a priority. an electronic nose (e-nose), with seven mos sensors, was used to analyze 53 samples of six different commercial instant coffees produced by the same industry. thereafter, chemometric tools such as common dimension analysis (comdim) and linear discriminant analysis (lda) were applied to classify the samples. comdim is an unsupervised multiblock analysis able to reduce large data dimensions from different tables. a block for each e-nose sensor with the first derivative of the transient signal was used as comdim input. four common dimensions (cds) were necessary to represent the e-nose data, which accumulated a total variance of 99.86%. salience tables indicate a relation in cd1 between sensors s1, s3, s5, s6, and s8. sensors s7 and s9 have more influence on cd2. the scores from the first four cds were used as input to construct lda classifiers. all models reached a sensitivity and specificity of 100% in the leave-one-out cross-validation. thus, the proposed approach was able to classify correctly the aromatic pattern of different commercial instant coffees.