Title is misleading. The article only describes univariate analysis (screening of independent variables).
One can build a model to predict churn from those independent variables (e.g., by just adding them up), but then there's out-of-time and out-of-sample validation, monitoring, and other boring, but important things.
And we like the "boring" things. But I don't see why doing univariate analysis doesn't deliver on the promise of the post. Contrasted with guessing, that is.
Online material on this subject is scattered, but there are good books like Regression Modeling Strategies by Frank E. Harrell and Credit Risk Scorecards by Naeem Siddiqi.
