I have got the following question:
I want to build a basic model which predicts whether a customer will churn or not using logistic regression to target 'high-risk' customers with, for instance, marketing initiatives.
For this, I have a dataset which contains information on past customers who have churned and those who did not churn (yet).
Intuitively, and after having 'trained' the model, I would assume that every customer who currently is still a customer but has been classified by the model as 'churned' to be a 'high-risk' customer.
Would this assumption be correct?
And: Assuming the model gets more and more data and becomes better and better. Would that not mean it would become worse and worse at detecting 'high-risk' customers as it would instead 'correctly' predict that a customer has not churned yet?
Much obliged for your help on this!