Brewston --Find your beer in any boston bars

My Insight data science project, which was done within 3 weeks time at Insight Boston

Slides:

https://docs.google.com/presentation/d/1Glii9f5FqkTYCsX49IthPRIASuCGOdfMqKKvAzYAx-I/edit#slide=id.g4c93729644_2_24

Webapp:

http://insight-app-yanchu.herokuapp.com/

Structure

• beeradv_crawler ---- a web crawler I wrote to scrape beer ratings & reviews from beeradvocate.com
  • brcrawl.py ---- crawler
  • brparse.py ---- parser 1
  • beer_brew_style_idmapping.py ---- parser 2
• beermenus_scraper ---- scraping scripts for beermenus.com, using selenium
  • bmscrp.py ---- scraper
  • bmpars.py ---- parser
  • matcher_google.py ---- script to match beer names between beermenus and beeradvocate using google search
• EDA.ipynb ---- show the data
• review_nlp
  • tfidf_glda.py ---- beer review topic modeling using guilded LDA with TFIDF
• recommendor
  • recomdl.py ---- recommendation model, with modification on the surprise package
• Heroku ---- Dash webapp running on heroku
  • app.py ---- Dash main file
  • utils
    • myweb.py ---- webpage setup
    • beername_matcher.py ---- match input beername with beeradvocate data, using TFIDF vector matching