Trump and the Art of Machine Learning!

Livestream link (https://livestream.com/h2oai/trumpml)

In this series of meetups till November Elections, we will analyze corpus of Trump's twitter feeds, talks and other public data sets and overlay them with polls. We will review the Times and other popular pollsters methods and techniques for predicting our future.

The repository for this meetup, including code demos, is on GitHub here (https://github.com/h2oai/h2o-meetups/tree/master/2016_10_06_TrumpTweets_Meetup).

Speaker Team: Avkash (Community), Data Science folks from H2O.

A hack-a-thon will be done as a prep for the event at 4pm and Community is welcome to join.

Agenda:

4:00 – 4:30 – Prep

4:30 – 5:00

• Feature Engineering, Text Analysis - Jay Navdeep

◦ How to convert raw tweet data to feature rich data ready for ML

5:00 – 8:15 - Hackathon

8:15 – 8:30 – Demo

• Sentiment Analysis - Erin

◦ Use dplyr, tidytext and rsparkling to analyze Trump tweets (iPhone (staff) vs Android (Trump))

◦ Perform sentiment analysis to find which are his own vs staff

8:30 – 8:45 - Demo

• Binary Classification - Wen, Surekha, Avni

◦ Question: Classify the tweet from either Hillary or Trump

◦ Question: Check If a tweet will be responded by Trump

8:45 – 9:00 – Demo & Wrap up

• Building & deploying machine learning applications - Avkash

◦ Export models from above classification exercise and build Java application using POJO