Tourism-Package-Purchase-RandomForest

Tourism Package Purchase Prediction

Model Description

This repository contains a trained Random Forest classification model developed to predict whether a customer will purchase a tourism package.

The model was developed as part of an end-to-end machine learning and MLOps project using the tourism customer dataset.

Model Details

• Model: Random Forest Classifier
• Task: Binary Classification
• Target Variable: ProdTaken
• Training Observations: 3,302
• Number of Features: 28
• Cross-Validation: 5-fold Stratified Cross-Validation
• Primary Model Selection Metric: ROC-AUC

Selected Hyperparameters

The final Random Forest model uses the following configuration:

• n_estimators: 200
• max_depth: None
• min_samples_split: 2
• min_samples_leaf: 1
• class_weight: balanced
• random_state: 42

Model Development

Random Forest, AdaBoost, and Gradient Boosting were evaluated during model development. Hyperparameter tuning was performed using GridSearchCV with five-fold stratified cross-validation.

A total of 49 hyperparameter configurations were evaluated across the three models.

The Random Forest configuration was selected based on the predefined ROC-AUC criterion.

Performance

Cross-Validation Performance

The selected Random Forest configuration achieved:

• Accuracy: 0.9013
• Precision: 0.9162
• Recall: 0.5392
• F1 Score: 0.6775
• ROC-AUC: 0.9612

Final Test Performance

The final model was evaluated on the previously unseen test dataset:

• Accuracy: 0.9080
• Precision: 0.9462
• Recall: 0.5535
• F1 Score: 0.6984
• ROC-AUC: 0.9681

Confusion Matrix

The final test-set confusion matrix was:

[[662   5]
 [ 71  88]]