According to the experimental setup, the perfect number of
groups that could be differentiated by the e-Nose is eleven and this
was associated to the 11 fruit juices. Other possible discrimination
groups suggest a cluster of the data into six (mango juices, pear
juices, orange juices, apple juice, pineapple juice and multivitamins
juice). As one of the goals of this work was to check the ability of
the e-Nose to distinguish all juices even if they are belonged to
same brand or had the same origin of fruit, the second option
was omitted. To apply PCA analysis, only two features extracted
from the response curve of each sensor (e.g., the steady-state con-
ductance and the dynamic slope of the conductance) were used.
Fig. 2 shows the three-dimensional representation of the three