🏦 Bank-churn-classification

🔗 Link - (## 🔗 Link - (https://www.kaggle.com/code/arkabarua173/bank-churn-xgboost-roc-89-shap?rvi=1)

• Please go to link to see the interactive plotly graphs.

📌 Objective:

• Predicting the probabilities of exiting the bank.

📄 Dataset Description:

• CustomerId: This is a unique identifier for each customer.
• Surname: The surname of each customer. Nominal variable.
• CreditScore: The credit score of each customer. Continuous Variable.
• Geography: Indicates the country where the customer resides. In this dataset (Germany, Spain & France).
• Gender: The gender of each customer.
• Age: The age of each customer. Continuous Variable.
• Tenure: The number of years the customer has been with the bank. Continuous Variable.
• Balance: The account balance of each customer. Continuous Variable.
• NumOfProducts: The number of bank products the customer uses.
• HasCrCard: Indicates whether the customer has a credit card (1 for yes, 0 for no).
• IsActiveMember: Specifies whether the customer is an active member (1 for yes, 0 for no).
• EstimatedSalary: Represents the estimated salary of the customer. Continuous Variable.
• Exited: This is the target variable and indicates whether the customer has exited the bank (1 for yes, 0 for no).

🔨 Data Preprocessing

• Mutual Information is used for selecting features.
• One Hot Encoder used for categorical value (In this dataset (Gender & Geography))

🤖 Model:

• XGBoost is used for this binary classification because dataset is imbalanced.
• XGBoost tends to work well with imbalanced dataset i.e. "scale_pos_weight" parameter can be utilized to handle imbalance.
• RandomizedSearchCV is used with 5 fold and roc_auc score for hypertuning.

🔍 Model Evaluation:

• Accuracy, Precision, F1 scroe, Recall and Area Under the Curve Receiver Operating Characteristic score is used for the model
• Best ROC_AUC score is 0.891.
• Class wise result are shown below:

• Confusion matrix are shown below

• Feature Importance based on weight, cover and gain are shown below

• Shap was used to understand the model output. Based on mean absolute Shapley values, NumOfProducts is the important feature followed by Age.