How can big data help make the perfect beer?

BY DENNIS HORTON

Big data and beer aren’t words that one would usually hear in the same sentence! But last year D2D CRC was lucky enough to work on a project where these words were routinely combined.

As part of the SA State Government’s Big Data Connect Program, D2D CRC worked with Coopers Brewery using big data analytics to perfect their flagship pale ale beer.

The eight-week project was led by D2D CRC Lead Data Scientist, Dennis Horton, as well as two D2D CRC data scientists.

What is the Big Data Connect Program?

The Big Data Connect Program is a South Australian state government funded initiative, which provides SA manufacturers with opportunities to use big data analytics, improving processes and getting the most out of their data.  

But how can big data help create the perfect beer?

There are all sorts of factors which make up the ‘perfect beer’, not least of which is personal taste and preference. So, the problem that we set out to solve with Coopers was how to improve the final quality of the product, including the formation and stability of the head of Coopers Pale Ale.

What did this involve?

We spent eight weeks developing an algorithm to perfect the brewing process and ensure a quality final product, including firmer foam that lingers longer.

The brewing process is complex and there can be variation in the raw ingredients, so it can be tricky to know which components have the main impact on product quality. But using data routinely collected by Coopers and applying data analysis, we were able to formulate valuable insights which will help ensure high quality outputs.

This involved linking and analysing data from various stages of the brewing process. Machine learning methods were investigated in order to use ingredient characteristics, process settings and other factors to predict final beer quality. We built a data-interrogation model that could identify important factors for the final quality of the beer based on the variability in the characteristics of the raw ingredients and process settings.

The algorithm identified the top ten most important components out of the hundreds of possible  brew settings and ingredients. These results provide valuable insights into the brewing process and can help Coopers manage the quality of the final product despite the complex nature of brew settings and ingredients.

So where to from here?

This project provided Coopers Brewery with valuable insights into the complex brewing process. The results both confirmed their previous ideas of the important components as well as revealed other factors that impact on final quality. These results warrant a follow up by the continuous improvement team at Coopers Brewery, who will assess the model for potentially adopting into the brewing process.