Skip to content
Jenna edited this page Apr 10, 2016 · 9 revisions

Idea

  • A fun little tool that will allow people to enter in a beer they like, and find other beers that match those characteristics (so they are likely to also enjoy)
  • Help people find new brewers/beer flavours they may not know about
  • Use similarity/distance approach to scoring beer similarities based on a feature vector that describes various characteristics (do analysis in R, maybe use Shiny or another tool to make an interactive chart)
  • This may be a whole big thing in itself, but brewing beer uses a lot of resources, namely water and chemicals. It would be cool to try to find out the “water footprint” of the microbreweries. This doesn’t really relate to project, but is something I’m curious about.

To Do

  • Come up with Attributes/Values
  • stouts, cream ales and porters are very different in terms of colour and taste to ales and lagers. I will try to think of markers for them. I guess a stout would be creamy/velvety, low carbonation, dark for now. Often stouts/porters are coffee or chocolate falvoured. [fun fact, the only difference between st ambroise cream and pale ale is cream all has nitrogen added, but the beers taste completely different]
  • We could look at type of beer, ie IPA, ESB, blonde, etc. but this might also take the fun out of the idea. If people generally go for IPAs, it might be better to not classify by type and try to recommend some different styles.

  • Add descriptors from key (ie color.1 --> pale yellow) back to page in Shiny app

1. Attributes brainstorming

Attribute Value
Colour Dark, light, yellow, amber, brown, cloudy, clear… [unfiltered vs filtered]
Taste light/crisp, bitter, fruity, sweet, hoppy, citron, sour, malty (sweet, caramel)
Body Rich, full, light, slick, creamy, oily, heavy, velvety, sweet, dry, thick, thin...
Carbonation Soft, effervescent, bubbly, gentle, low carbonation, highly carbonated...
Hops used Two main types of hops: bittering and aromatic. The hop itself also has an acidity index. The higher the alpha-acid index, the greater the bitterness. We could just get brewers to estimate bitterness of hops on a numerical scale, and then maybe think of a way to classify aromatic hops.
Alcohol content %

2. Come up w/ list of Toronto brewer's and contact info

3. Data Gathering

  • Google forms → feeds into google spreadsheets
  • Send out form to brewers where they are asked to list the beer name, and the attributes about it
  • Free text vs dropdowns?
  • Beers may have multiple values for each attribute
  • Make survey for brewers that they can tick off different flavours/attributes for each beer

4. Data Analysis

  • Similarity/distance measuring
  • Use weighting for attributes with varying number of value options?

5. Visualization/tool

  • Use Shiny/other visualization tool to allow user to enter or choose a beer they like, and then it will find the beer’s “nearest neighbours”

Resources/Helpful things