From the course: Executive Guide to Predictive Modeling Strategy at Scale
Join today to access over 22,600 courses taught by industry experts or purchase this course individually.
Combining batch and real-time scoring
From the course: Executive Guide to Predictive Modeling Strategy at Scale
Combining batch and real-time scoring
- [Instructor] Now let's revisit the call center example and see if there are circumstances under which we would definitely want real-time scoring. For instance, let's say someone's calling in around a cell phone repair, or possibly even an upgrade, and they've had three complaints over the last seven days. It would certainly seem inappropriate to not take that recent activity into account if we can, so what would be the most efficient way to do so? This is what people forget. We can actually do a mix of batch and real-time. We can do our standard model, which is taking our billing data only updated monthly, we can run it through our model, and we generate our churn score. Keep in mind that this model is using dozens of variables. The data prep takes several minutes to run. It's run in batch. The system's already there. We want to leverage this, not rebuild it, so this is what we do. We take that churn score from the first model, and we simply add to it only those things that change…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.