TelecomGiant is losing customers at an alarming rate
It's 6:00 AM on a Tuesday. You're the newly hired Head of Data Science at TelecomGiant, one of the largest telecom providers in North America with 12 million subscribers.
The CEO bursts into your office: "We're hemorrhaging customers! Last quarter alone, we lost 850,000 subscribers. That's $45 million in annual recurring revenue - GONE!"
Your predecessor tried to predict a continuous value (customer lifetime value), but what we really need to know is binary: Will they churn? Yes or No.
What type of problem are we solving?
Logistic regression transforms linear combinations into probabilities using the sigmoid function:
The sigmoid function ensures our predictions stay between 0 and 1 - perfect for probabilities!
Adjust the classification threshold and see how it affects our business metrics:
The cost of missing a churner is 9x higher than unnecessary retention!