Model Card: Gradient Boosting

Model Description

This model is a Gradient Boosting Classifier trained to predict tourism package purchases. It has been tuned using GridSearchCV on a SMOTE-resampled training dataset to address class imbalance.

Tuned Parameters

Parameters: {'learning_rate': 0.2, 'max_depth': 7, 'n_estimators': 300}

Evaluation Metrics (on Test Set)

*   Accuracy: 0.9467
*   Precision: 0.9389
*   Recall: 0.7736
*   F1-Score: 0.8483
*   ROC AUC: 0.9771

Model Purpose

This model aims to accurately predict whether a customer will take a tourism package, aiding in targeted marketing efforts and improving sales efficiency.

Usage

To use this model, load the gradient_boosting_model.pkl file and apply it to preprocessed data. Ensure the data undergoes the same preprocessing steps (column removal, outlier handling, one-hot encoding, and StandardScaler scaling) as the training data.