For the considered samples the classification capability of ANN and e-nose coupling were higher than the technique based on GC– MS analysis of volatile compounds.
The results have shown how the ANN and e-nose coupling can be a very effective classification technique for Pecorino cheese, easy to implement and not computationally expensive, even if needs to be well understood and set up by researcher.
It is always possible to consider the entire dataset, accepting an higher computational effort, even if the small loss of information due to PCs extraction make possible to reduce calculation time without significant precision loss. On the contrary, data feature extraction was not a successfully technique.