Customer Attrition Model

[Ujjawal]

I am in the process of building a customer attrition model on retail customers of saving account for a bank. The target variable i am defining consists of two components - 1. Hard Attrition : Bank loses the account completely. 2. 80% balance drop within the performance period of last 12 months. My question - Can i take balance variables as independent variables? By the balance variables, i mean - Last Balance, Average balance in 3 months,  in 6 months,  9 months and 12 months. %Change Balance in last 3 months, 6 months, 9 and 12 months. My concern - These variables indirectly seem to be a part of dependent variable. Am i correct? Or i am thinking in wrong direction? I have seen these balance variables as important predictors in the model.

[Reeza]

The balance variables will by definition be related to the variable so in my opinion they won't be good predictors. Obviously if someone has a major drop they're likely to be in the 80% group...that being said, time series analysis does allow for this type of modelling so you may want to look into some time series methods of analysis.

[JasonXin]

Hi, 1. If you are into forecasting mind set, yes you can retain those variables in your model. We could call them autoregression. 2. Back to linear regression kind of predictive models, a 'taboo' in practice is NOT to include any variables used in defining the target. in most cases it means variables with the same names.Sometimes it refers essentially the same variables with different names. 3. After excluding these variables, variables, as described in the original questions, can be highly correlated to the defined target. That per se is not problem. Another subject is they are highly correlated among themselves. You can either pick one of them or load them into factors or components. Hope this helps? Jason Xin