Youngkin V. McAuliffe: RealClearPolitics.com Vs Tweets
Find full code on my Github.
Glenn Younkin(R) pulls ahead in the poll (Real Clear Politics) on 10/23. On Twitter, Younkin pulls ahead on 10/25.
- Both show a surge beginning 5/23.
The Real Clear Politics(RCP) graph is made up of the aggregation of polls from the RCP site. A sigmoid function was used to bring all the values first between 0and 1, and then multiplied to fit between -1 and 1.
The Twitter data is from Twitter users who claim to be from VA in their profile. Unfortunately, 2 days of data (10/29 & 10/30) were not collected due to connection issues. On those days the graphs show no change in the tweet data.
The Twitter graph is a moving average of 3days.
Over 600,000 tweets were collected and of that 2,000 were from 'VA residents.'
Error: How Close Were these graph lines "technically"
(not really very close)
You want RMSE to be close to 0 and R-Squared to be close to 1 so....ya know.