# Churn Prediction 'Bank of Gardenia' Hackathon

#1

What is interpreted by Responder (target) variable as 1 or 0? Does 1 mean customer has churned out or opened a savings account?? The question is not clear. What are we supposed to predict?? Please help here.

#2

Same Question is the responder the Target variable ? as the question was predict for customers whose worth reduces by 50 % and one more sayng opt for the savings account . If some one can explain the problem statement a bit better it would be helpful.

#3

In one statement it is written that - ‘BoG wants you to identify customers likely to churn balances in the next quarter by atleast 50% vis-a-vis current quarter’.

‘Estimate the probability of customer to opt savings account’ - and also this also written.

So what we have to predict ? - Customers withdrawing 50% of their balance in next year or customer closing their Savings account?

#4

Kindly respond. whether Responder is target variable & What 0 & 1 stand for?

#5

Well I was also confused with the problem statement, but this is what I’ve interpreted so far, and I’m welcome to new ideas -

1. The customer base given in train will be as of a certain date, and comprises of all customers (churned / new / existing etc.)
2. Responders are the customers who have churned from the bank in some period, and are tagged by 1.
3. Its not necessary that the responders will have declining AQB’s by 50% in two consecutive quarters following the churn. Hence, we see flag of 1 for some customers, whose balances have increased as well.
4. The problem statement here is to predict customers who will be churning / or declining their AQB’s by 50% from current to next quarter. So it might be a good idea to create a separate response variables - one for AQB, and one for churn, and predict them separately!

#6

I have updated the problem to clarify the confusion. The problem statement is to identify customers likely to churn balances in the next quarter by atleast 50% vis-a-vis current quarter