Brewers all over the globe are embracing artificial intelligence and machine learning as tools to make their processes more effective and to produce targeted data to enhance their brewing. Champion Brewing in Charlottesville in Virginiarecently teamed up with Metis Machine, a machine learning company, to try and discover the recipe for the perfect IPA.
By using data on which IPAs were the most and least popular with consumers Metis created a machine learning model which could then determine the probable success of a beer based on its IBU (International Bitterness Units), ABV (Alcohol By Volume), and SRM (Standard Reference Method for classifying colour).
The team began by collating data on the most and least popular IPAs on the market. Hunter Smith, owner of Champion Brewing Company, explained that they used “the parameters on which IPAs are judged at the Great American Beer Festival (SRM, ABV, IBU) and matched that range with the 10-best-selling IPAs nationally, as well as the 10 worst selling IPAs at a local retailer and Metis came up with the results.”
Once the data had been generated, Hunter Smith was able to create a recipe based on the SRM, ABV, and IBU recommended by the machine learning model. The fruit of the project was ML IPA which has 6% ABV, 60 IBU, and 6 SRM which falls more or less in the middle of these parameters.
Metis Machine’s data science engineer, Tyler Hutcherson, explained that the company’s aims during the project were “to demonstrate that artificial intelligence, while perceived to be quite daunting, can lend itself to many practical domains, including craft brewing.”