| Overview: |
This talk will explore the use of predictive analytics to classify users according to their propensity to perform an action of interest in the near future, based on their interactions with a game. Prediction alone is not enough, one needs to take some action based on the prediction. The talk will also explore how a system may be created, in the context of an online game, to predict behavior, score users, test different actions, and optimize strategies for exploiting the results of the prediction. The talk will evaluate logistic regression methods, and explore their use in predicting player behavior.
|
