48
What can teach us about @martinjc PyDiff - December 15th 2015

PyDiff - what python can teach us about beer

Embed Size (px)

Citation preview

Page 1: PyDiff - what python can teach us about beer

What can teach us about

@martinjc PyDiff - December 15th 2015

Page 2: PyDiff - what python can teach us about beer
Page 3: PyDiff - what python can teach us about beer
Page 4: PyDiff - what python can teach us about beer
Page 5: PyDiff - what python can teach us about beer

Untappd API available

‘local’ endpoint returns all beer checkins within set radius of given point

if they’re associated with a venue

Page 6: PyDiff - what python can teach us about beer

Dr Luca Rossi Aston University

@blextar

Dr Gareth Tyson Queen Mary University, London

@Gareth_Tyson

Dr Matthew J Williams Birmingham University

@voxmjw

Page 7: PyDiff - what python can teach us about beer
Page 8: PyDiff - what python can teach us about beer
Page 9: PyDiff - what python can teach us about beer

… live coding …

Page 10: PyDiff - what python can teach us about beer

data collection

monitoring 40 cities

34 USA, 6 EU

from 13th August 2015 onwards

Page 11: PyDiff - what python can teach us about beer

As of 12th December 2015:

5,770,121 checkins

made by 365,788 users

drinking 137,868 different beers

at 322,615 different venues

Page 12: PyDiff - what python can teach us about beer
Page 13: PyDiff - what python can teach us about beer
Page 14: PyDiff - what python can teach us about beer

Checkins1. Chicago 2. New York 3. Philadelphia 4. Washington DC 5. Denver 6. Portland 7. Boston 8. San Diego 9. Seattle 10. London

1. New York 2. Chicago 3. Denver 4. Philadelphia 5. London 6. Washington DC 7. Seattle 8. Portland 9. San Diego 10. San Francisco

Unique Beers

Page 15: PyDiff - what python can teach us about beer

… so where are the users with a high diversity in

beers?

Page 16: PyDiff - what python can teach us about beer
Page 17: PyDiff - what python can teach us about beer
Page 18: PyDiff - what python can teach us about beer

… who drinks the beers with the highest ABV?

Page 19: PyDiff - what python can teach us about beer
Page 20: PyDiff - what python can teach us about beer

… what is the most popular beer in the

world*?

*if the world is restricted to Untappd users within 40 cities in the US and Europe…

Page 21: PyDiff - what python can teach us about beer

1 (21605): Boston Beer Company (Samuel Adams) - Samuel Adams OctoberFest (5.30%) - United States (MA) 2 (19899): Lagunitas Brewing Company - IPA (6.20%) - United States (CA) 3 (19066): Guinness - Guinness Draught (4.30%) - Ireland (Ireland) 4 (17601): Goose Island Beer Co. - Goose IPA (5.90%) - United States (IL) 5 (15945): Anheuser-Busch - Bud Light (4.20%) - United States (MO) 6 (14795): Miller Brewing Company - Miller Lite (4.17%) - United States (WI) 7 (13527): Ballast Point Brewing & Spirits - Sculpin IPA (7.00%) - United States (CA) 8 (12724): Bell's Brewery, Inc. - Two Hearted Ale (7.00%) - United States (MI) 9 (12304): Ballast Point Brewing & Spirits - Grapefruit Sculpin (7.00%) - United States (CA) 10 (11772): Coors Brewing Company - Coors Light (4.20%) - United States (CO)

Page 22: PyDiff - what python can teach us about beer

… what is the most popular beer in europe*?

*if Europe is restricted to Untappd users within 6 cities…

Page 23: PyDiff - what python can teach us about beer

1 (7026): Heineken - Heineken (5.00%) - Netherlands (Zuid Holland) 2 (5804): Guinness - Guinness Draught (4.30%) - Ireland (Ireland) 3 (2610): BrewDog - Punk IPA (5.60%) - Scotland (Aberdeenshire, AB41 8BX) 4 (2061): Meantime Brewing Company - London Pale Ale (4.30%) - England (Greater London, SE10 0AR) 5 (1944): Fuller, Smith & Turner - London Pride (4.10%) - England () 6 (1731): Brasserie d'Achouffe - La Chouffe (8.00%) - Belgium () 7 (1644): Amstel Brouwerij - Amstel (5.00%) - Netherlands (Zuid Holland) 8 (1628): Grupo Damm - Estrella Damm (5.40%) - Spain (Cataluña) 9 (1559): Koninklijke Grolsch - Premium Pilsner / Premium Lager (5.00%) - Netherlands (Overijssel) 10 (1536): Carlsberg Group - Carlsberg (5.00%) - Denmark ()

Page 24: PyDiff - what python can teach us about beer

… where does the beer come from?

Page 25: PyDiff - what python can teach us about beer

Los Angeles

1275: Ballast Point Brewing & Spirits - Sculpin IPA (7.00%) - United States (CA) 1231: Angel City Brewery - Angel City IPA (6.10%) - United States (CA) 1188: Golden Road Brewing - Point The Way IPA (5.90%) - United States (CA) 1084: Russian River Brewing Company - Pliny the Elder (8.00%) - United States (CA) 966: Lagunitas Brewing Company - IPA (6.20%) - United States (CA)

Page 26: PyDiff - what python can teach us about beer

Baltimore

2247: Heavy Seas Beer - Loose Cannon (7.25%) - United States (MD) 1973: Pabst Brewing Company - National Bohemian (4.60%) - United States (IL) 1428: Boston Beer Company (Samuel Adams) - Samuel Adams OctoberFest (5.30%) - United States (MA) 995: Union Craft Brewing - Duckpin Pale Ale (5.50%) - United States (MD) 985: The Brewer's Art - Resurrection (7.00%) - United States (MD)

Page 27: PyDiff - what python can teach us about beer

Detroit

2011: Bell's Brewery, Inc. - Two Hearted Ale (7.00%) - United States (MI) 1816: Griffin Claw Brewing Company - Norm's Raggedy Ass IPA (7.20%) - United States (MI) 971: Boston Beer Company (Samuel Adams) - Samuel Adams OctoberFest (5.30%) - United States (MA) 908: Griffin Claw Brewing Company - Screamin' Pumpkin (5.00%) - United States (MI) 897: Perrin Brewing Company - Blackberry IPA (6.00%) - United States (MI)

Page 28: PyDiff - what python can teach us about beer

Copenhagen

746: Carlsberg Group - Tuborg Classic (4.60%) - Denmark () 640: Carlsberg Group - Carlsberg (5.00%) - Denmark () 441: Carlsberg Group - Tuborg Julebryg (5.60%) - Denmark () 432: Carlsberg Group - Brewmasters Collection India Pale Ale (5.20%) - Denmark () 398: Carlsberg Group - Tuborg Grøn (4.60%) - Denmark ()

Page 29: PyDiff - what python can teach us about beer

… and when do people drink it?

Page 30: PyDiff - what python can teach us about beer
Page 31: PyDiff - what python can teach us about beer

… what styles do people drink?

Page 32: PyDiff - what python can teach us about beer
Page 33: PyDiff - what python can teach us about beer
Page 34: PyDiff - what python can teach us about beer
Page 35: PyDiff - what python can teach us about beer

… are the more popular beers rated highly?

Page 36: PyDiff - what python can teach us about beer
Page 37: PyDiff - what python can teach us about beer

… are there other differences between the

US and the EU?

Page 38: PyDiff - what python can teach us about beer
Page 39: PyDiff - what python can teach us about beer
Page 40: PyDiff - what python can teach us about beer

Untappd• Untappd is fairly popular

• Particularly large US cities

• (Chicago, New York)

• Popularity in EU variable

• (London high, Paris low)

Page 41: PyDiff - what python can teach us about beer

Beer Diversity• High numbers of beers in city != high user diversity

• Diversity = abundance + variety

• Spikes + uneveness = low diversity

• London: high beers, low user diversity

• San Jose: lower beers: high user diversity

Page 42: PyDiff - what python can teach us about beer

Beer Strength

• London & Dublin - lower beer strength than elsewhere

• California higher

Page 43: PyDiff - what python can teach us about beer

Popularity

• Global popularity = big brands & breweries

• Local popularity = smaller local breweries

Page 44: PyDiff - what python can teach us about beer

Time

• Beer is mainly drunk in the evenings

• Except at the weekend!

• Or in London!

• Beer is not more popular on the weekend in Paris

Page 45: PyDiff - what python can teach us about beer

Styles

• Globally: ales win, then lagers

• Locally: variable

• Often ‘local’ ales are more popular

Page 46: PyDiff - what python can teach us about beer

Ratings

• Beers with more checkins tend to have a higher rating

• US are more generous than EU with ratings

• or is US beer better? (:-O)

Page 47: PyDiff - what python can teach us about beer

What next?• Social Network analysis

• homophily

• beer spread (epidemic?)

• network growth

• Binge drinking

• Beer festivals/event detection

Page 48: PyDiff - what python can teach us about beer

bardiff.martinjc.com