problem with type 1 error while using SMOTE and XGBoost on imbalanced dataset

Question

I\'m recently working on a dataset that\'s highly skewed. while making a predictive model for accepting or rejecting new customers to join the contract.

There are app