- This event has passed.
Why Building Model Ensembles is a Game Changer
May 26, 2017Free
Session Time and Date
Friday, May 26th from 3pm – 4pm CDT
How To Attend A Live Conference Session
Steps to Follow
- Click the link below to attend:
- Name and Email
- Join the conference session by clicking the link on the GoToWebinar confirmation email
The most effective approach to win predictive analytics data competitions and producing highly accurate predictive models is the use of model ensembles, a technique that combines predictions from multiple models into a single score. The use of ensembles has revolutionized predictive modeling not just in competitions, such as the Netflix Prize, Kaggle, and KDD CUP competitions, but also in everyday modeling for private and public sector organizations. This talk introduces model ensembles and will walk through the history of model ensembles in machine learning and predictive analytics, including Bagging, Boosting, Random Forests, Stochastic Gradient Boosting, and heterogeneous ensembles. While ensembles appear to be more complex than individual models, thus violating Occams Razor, this talk will also unravel the apparent contradiction. Real-world examples of the application of model ensembles will be provided throughout the talk.