Use and misuse of supervised pattern recognition methods for interpreting compositional data

In this paper, the critical aspects of supervised learning pattern recognition techniques are examined in order to apply the suitable methodology avoiding its abuse and misuse that could provide pitfalls and misleading results.