To predict telecom churn(Xius)
Using some important fields like Plan.code,call summary,data usage,days.since.last.usage,customer age ,I created a binary classification problem with the target variable as "customer status". Here customer status: Active - means customer is using our services. customer status: suspended - means customer is left from the service. Note:the minority class is poor in this data (<1%) ,so here got ~99% Accuracy .It means accuracy is misleading due to algorithm is neglecting minority class(suspended). So need to use sampling techniques for populating the minority class.