In an embodiment, in building predictive models, rmm_analytics schema are filtered based on the flow of historical leads through the inbound call center routing process. In an embodiment, the data are filtered to select only historical leads that were connected to a live agent; in the present disclosure this flow is sometimes called a “warm transfer.” Applicant has observed that building predictive models based on a population limited to warm transfers can improve performance of models for predicting sales and lapse behaviors.
In the illustrative embodiment, data used in predictive modeling also include data retrieved from the customer demographic database 432 to obtain information about customers. In an embodiment, customer demographic data includes individual level data on customers. In various embodiments, as a prerequisite to using data in predictive modeling of a given inbound caller (customer), analytical engine 404 associates the customer demographic data with a customer identifier for the customer. In an illustrative embodiment, customer demographic data used in modeling of a customer requires an exact match of name and address.