Fig. 2: Unsupervised classification of meat flosses using principal component analysis (PCA) integrated with time window slicing method.

PCA is implemented to analyze output sensing signals that are preprocessed with four different extracted features: a maximum, b minimum, c mean, and d median values. Time window slicing method is applied to construct different window numbers in the data (i.e., 1 window (W1) and 5 windows (W5)). The condition without window (W0) is also analyzed as reference. PCA can create separated clusters between pork and non-pork meat flosses in the data measured by e-nose, despite the existing overlap between beef and chicken classes.