PREDICTING WINNERS OF THE BIG GAME
WITH WSO2 MACHINE LEARNER
Nirmal FernandoTeam Lead of WSO2 Machine Learner
What is the Big Game?
http://news.yahoo.com/american-football-broncos-reach-super-bowl-edging-patriots-010504321.html
Why were we interested in it?“Here's a crazy idea...a really crazy idea in fact…”
What data do we have? - offensive strength
By Ben Stanfield (user name acaben) of flickr - http://www.flickr.com/photos/acaben/1307980635/in/set-72157601826753840/, CC BY-SA 2.0, https://commons.wikimedia.org/w/index.php?curid=2689957
What data do we have? - defensive strength
http://bigeasybeliever.com/2015/03/31/saints-may-target-all-american-linebacker-with-31st-pick/
What type of a model do we build?
Winning probability
How we gathered the data
How we picked the Machine Learning algorithm
Response variable: Winning probability (numerical value)
Dataset: Supervised
Algorithm Type: Numerical Prediction
Algorithms supported by WSO2 Machine Learner
What we picked
The model we built
AWS
WSO2 ML Predictive API
How we predicted game results
Big Game REST API
Team A, Team B
Team A, Team B
Team A = 0.67 Team B = 0.33
Team A = 67% Team B = 33%
The final results
http://wso2.com/landing/big-data-game/
Challenges we faced
● Domain knowledge
● Lack of data
● Impact from injuries
● Game day condition
etc..
Interested in getting started with your own predictions?
● Product download http://wso2.com/products/machine-learner/● Product documentation https://docs.wso2.
com/display/ML110/WSO2+Machine+Learner+Documentation● Upcoming webinars http://wso2.com/events/webinars/
Thank You!
