Inspiration

Our conversations with the judges and mentors led us to focus on a key issue facing AB InBev (as well as other manufacturers): The lack of localized data and ultimately ability to provide actionable insights to merchants and distributors for micro-campaigns and endeavors to enhance the local beer drinking experience.

What it does

  • Collects user intent and purchase data via a chrome extension that enables beer and wine pairings for every entree on delivery.com
  • Collects user intent and purchase data via an Amazon Echo application that enables easy access to beer specific information and the ability to purchase beer vocally
  • Aggregates and displays user intent and purchase data empowering visualization of correlative behaviors
  • Appends third-party data sets (sporting events, weather) and overlays in real time events over user intent and purchase behavior to enable actionable, local insights and predictive modeling (e.g., basketball games in the city of Houston during cold days is highly correlated to Corona sales)
  • IBM-Watson integration to power predictive modeling and filtering data by time, product, merchant, beer and third-party data segments

How we built it

  • Chrome-extension (javasript, ajax)
  • Amazon Echo
  • Python/flask
  • Google App Engine
  • Angular.js
  • Delivery.com API
  • IBM-Watson

Challenges we ran into

  • Exploring delivery.com's site via chrome inspector (no 'documentation')
  • Chrome-extensions cannot post requests to non-https urls
  • Normalizing data to feed into IBM-Watson analytics (some of the metrics didn't graph correctly)
  • Real time data visualization (syncing the data collection with the analysis in real time)

Accomplishments that we're proud of

  • Beer and wine pairings for each entree on delivery.com's menus and integrating with delivery.com's API and our analysis engine
  • Complete end-to-end solution
  • Live data and real time updates to the dashboard
  • Populate and scale relatively well with google-app-engine (hundreds of thousands of transactions)
  • Merchant and product data aggregation via delivery.com
  • Delivery execution and transaction through Alexa voice command
  • Using IBM-Watson analytics to import data and provide predictive analytics

What we learned

  • Working together in a team, learning new skills (Google-app-engine for Eugene, Alexa/Angular.js for Saj), and creating a cohesive product among multiple sources of live and offline data

What's next for Parati

  • Additional live data sources
  • Scaling existing (e.g., getting the chrome-extension into hundreds of user's hands)
  • Appending additional third party data sets to overlay to find correlative behavior
  • Additional data sets for IBM-Watson
  • Expand current product/merchant directory via additional data sources and APIS (Untappd, beeradvocate, etc.)
  • Sentiment analysis for social media outlets (Twitter, FB, etc.)
  • Working with brewery experts to drive hypothesis driven analyses (adding specific data sets we as industry novices would not expect)
Share this project:
×

Updates