Algorithms can determine whether a whiskey is of American or Scotch origin

Two machine learning algorithms can determine whether a whiskey is of American or Scotch origin and identify its strongest aromas, according to research published in Communications Chemistry. The results also suggest that the algorithms can outperform human experts at assessing a whisky's strongest aromas.

A whiskey's aroma is determined by a complex mixture of odorous compounds. This makes it highly challenging to assess or predict a whiskey's aroma characteristics, or notes, based solely on its molecular composition.

Panels of human experts are often used to identify the strongest notes of a whiskey, but these require a significant investment in time, money, and training, and agreement between participants is often limited.

Andreas Grasskamp and colleagues assessed the molecular composition of seven American and nine Scotch whiskies using two algorithms—OWSum, a molecular odor prediction algorithm developed by the authors, and a neural network.

The molecular composition data was derived from existing results from gas chromatography and mass spectrometry analysis—two techniques used to separate and identify components within a mixture. The algorithms were used to identify each whisky's country of origin and its five strongest notes. The authors then compared the algorithms' results to those from a panel of 11 experts.

OWSum was able to determine whether a whiskey was American or Scotch with a greater than 90% accuracy.

Detection of the compounds menthol and citronellol was most closely associated with an American classification, while detection of methyl decanoate and heptanoic acid was most closely associated with a classification as Scotch.

OWSum identified caramel-like as the most characteristic note of American whiskies, and apple-like, solvent-like, and phenolic (often described as a smoky or medicinal smell) as the most characteristic notes of Scotch whiskies. Finally, both algorithms were able to identify the five strongest notes of a specific whiskey more accurately and consistently on average than any individual human expert.

The authors believe that their approach could lead to quick algorithmic classification of whiskies and identification of the key notes in their aromas.