Developed logistic regression, random forest and XGBoost models with bank marketing data (20 features and 41,000 records) where the optimal XGBoost model reached an AUC of 81%.
Imcreased the model performance by feature engineering like removing irrelevant and high correlation features and one-hot encoding for categorical features.
Set three differernt metrics of model performance and data's distribution to monitor when to retrained the model.
Deployed the model by creating a Web API which receives client data through GUI and returns the predicted result with Flask.