What are people drinking this season?
Homebrewers, craft brewers, and even larger brewers have all run into this question one time or another. There are, of course, your mainstays: Pumpkin Ale and Octoberfest for October, Winter Ales and Porters for the winter, Hefeweisen for summer, and so on. The selection of beers at your typical store changes based on the season all the time. This is what we call normal data; stuff we already know.
But what other beers are consumers looking for? What're people drinking that we don't know about, and when are they drinking it? What don't we know?
Those are the questions I attempted to (at least start to) answer in a recent analysis of a dataset including over 1.5 million beer reviews, over 400,000 contributors, and 12 (2000-2012) years of data.
Beer as Big Business
Craft beer, and craft breweries, are a growing industry that deserve more analysis. Almost 200 million barrels (bbls), or six billion gallons of beer, were sold in 2013. Of that, over 15 million barrels, or nearly 500 million gallons, were craft beer. $100 billion dollars was made on beer last year, and of that, $14.3 billion was spent on craft beers. This is a 17.2% increase in craft beer sales from 2012-2013.1
Craft beer is now big business, and because of that growth, larger breweries are getting into the game, too. Yet despite all the growth and buzz surrounding craft beer, it's still a relatively nascient market. So when we ask questions like "what beer should we brew?" the answers aren't always available.
This is a solveable problem.
There's tons of great beer data already out there, often going back over a decade. Untappd has an API to access their data, for instance. Additionally, Twitter and Facebook are treasure-troves of people posting what they're drinking almost religiously. Even review sites like Beer Advocate have, from time-to-time, opened up their data. We can, then, collect that data and analyze it using typical visualization and insight methodologies. From there, we can infer what beers we should brew.
So, with a few caveats, I present my first small crack at the data heap I've got at my fingertips. At its core, the visualizations below represent a way to see what beers styles and brands are popular in any given season.
- Things like IPAs and Pale Ales have been filtered out of the bubble graph because, frankly, they pollute the data (they're constantly popular -- come on, people). However, you can still find IPA data in the lower graph.
- By hovering over the bar graph in the top right corner, you can see what beer brands are being reviewed frequently. Some of this data was also normalized to provide a level of accuracy.
- I'll be updating this page over the coming weeks. Probably with some more complex D3 visualizations. In the process, I'll be adding the following: word clouds for flavor profiles, Twitter and Untappd data visualizations, regionalized visualizations, more normalized data, aggregated data combined with beer ratings, and sentiment analyses of beer brands and types.
Given those caveats, I still believe this data is an interesting start and gives us all something to think about. Enjoy: