|Session Name:||Predicting Player Behaviors: Lessons from TelCo's & Finance|
|Track / Format:||Business and Marketing|
Telecommunications and financial services have mined customer-behavior data to derive predictions, identifying the specific customers most likely to unsubscribe (churn) or to purchase added services (upsell). Social & online games also generate rich data that is often suitable for predictions of conversion, churn, and item purchase. This lecture presents case studies from all these industries, showing applications of predictive analytics, and including specific results. Predictives will be introduced, and contrasted to metrics. Potential uses will be suggested, from promotions to automated player-lifecycle management. A simple roll-your-own method will be described and commercial methods using unsupervised machine learning, neural networks, and social-graph analysis.