🌴 Holiday Package Prediction 🌴

With Random Forest Classification

---

📌 Project Overview

Trips & Travel.Com wants to expand its customer base by launching a new Wellness Tourism Package — defined as travel that allows the traveler to maintain, enhance or kick-start a healthy lifestyle and sense of well-being.

Previously, customers were contacted randomly for marketing without using any available data — leading to high marketing costs and a low conversion rate of only 18%.

This project builds a Random Forest Classification model to predict which customers are likely to purchase the Wellness Tourism Package, so the company can target the right customers and make marketing expenditure more efficient.

---

📁 Repository Structure

Holiday-Package-Prediction-Randomforest/
├── images/
│   ├── holiday_package_prediction.gif
│   ├── feature_distributions.png
│   ├── correlation_heatmap.png
│   ├── confusion_matrix.png
│   ├── feature_importance.png
│   └── auc.png
│
├── Holiday_Package_Prediction.ipynb
├── Travel.csv
├── holiday_package_classification_model.pkl
├── preprocessor.pkl
├── .gitignore
└── README.md

---

🛠️ Libraries Used

• scikit-learn
• pandas
• numpy
• matplotlib
• seaborn
• joblib

---

📊 Dataset

The dataset is sourced from Kaggle — Holiday Package Purchase Prediction

• Samples: 4,888
• Features: 20
• Target Variable: ProdTaken — 1 (Purchased) / 0 (Not Purchased)
• Purchase Rate: ~18% (class imbalance present)

Feature	Description
Age	Age of the customer
MonthlyIncome	Monthly income of the customer
DurationOfPitch	Duration of the sales pitch (in minutes)
Passport	Whether the customer has a passport (1/0)
NumberOfTrips	Number of trips taken by the customer
PitchSatisfactionScore	Customer satisfaction score for the pitch
NumberOfFollowups	Number of followups made by the agent
TotalVisiting	Total number of people visiting (person + children)
CityTier	Tier of the city the customer belongs to
Occupation	Customer's occupation

---

⚙️ Workflow

1. Load the dataset from Travel.csv
2. Handle missing values — median for continuous, mode for discrete features
3. Fix categorical inconsistencies (Fe Male → Female, Single → Unmarried)
4. Drop irrelevant feature — CustomerID
5. Feature Engineering — create TotalVisiting from NumberOfPersonVisiting + NumberOfChildrenVisiting
6. Exploratory Data Analysis — distributions and correlation heatmap
7. Train-Test Split — 80% train / 20% test (random_state=42)
8. Preprocessing — OneHotEncoder for categorical, StandardScaler for numerical via ColumnTransformer
9. Train and compare multiple models — Logistic Regression, Decision Tree, Random Forest, Gradient Boosting
10. Hyperparameter tuning using RandomizedSearchCV with 100 iterations and 3-fold CV
11. Evaluate using Confusion Matrix, Classification Report, and ROC-AUC Curve
12. Feature Importance analysis
13. Save model and preprocessor using joblib

---

📈 Exploratory Data Analysis

 

Key Insights:

• Passport has the strongest positive correlation with purchase (0.26)
• Age and MonthlyIncome are highly correlated with each other (0.46)
• DurationOfPitch and NumberOfFollowups positively influence purchase decision
• Most customers are from City Tier 1 and prefer 3-star properties

---

🤖 Model Comparison

Model	Train Accuracy	Test Accuracy
Logistic Regression	~81%	~81%
Decision Tree	~100%	~85%
Gradient Boosting	~93%	~90%
Random Forest	~100%	~93%

Random Forest was selected as the final model based on best test accuracy and F1 score.

---

🔧 Hyperparameter Tuning

rf_params = {
    "max_depth"        : [5, 8, 15, None, 10],
    "max_features"     : [5, 7, "sqrt", 8],
    "min_samples_split": [2, 8, 15, 20],
    "n_estimators"     : [100, 200, 500, 1000]
}

Tuning method: RandomizedSearchCV — n_iter=100, cv=3, n_jobs=-1

Best Parameters:

RandomForestClassifier(
    n_estimators=1000,
    min_samples_split=2,
    max_features=7,
    max_depth=None
)

---

📉 Model Results

 

Classification Report:

               precision    recall  f1-score   support

Not Purchased     0.93      0.99      0.96       787
    Purchased     0.97      0.68      0.80       191

     accuracy                         0.93       978
    macro avg     0.95      0.84      0.88       978
 weighted avg     0.94      0.93      0.93       978

Top Predictive Features:

1. 💰 MonthlyIncome
2. 🎂 Age
3. ⏱️ DurationOfPitch
4. 🛂 Passport
5. ✈️ NumberOfTrips

---

🧪 Sample Prediction

import joblib
import pandas as pd

model = joblib.load('holiday_package_classification_model.pkl')
preprocessor = joblib.load('preprocessor.pkl')

# New customer sample
sample = pd.DataFrame({
    'TypeofContact'         : ['Self Enquiry'],
    'CityTier'              : [1],
    'DurationOfPitch'       : [20],
    'Gender'                : ['Male'],
    'NumberOfFollowups'     : [4],
    'ProductPitched'        : ['Deluxe'],
    'PreferredPropertyStar' : [3],
    'MaritalStatus'         : ['Unmarried'],
    'NumberOfTrips'         : [3],
    'Passport'              : [1],
    'PitchSatisfactionScore': [4],
    'OwnCar'                : [1],
    'Occupation'            : ['Salaried'],
    'MonthlyIncome'         : [25000],
    'Age'                   : [30],
    'Designation'           : ['Executive'],
    'TotalVisiting'         : [3],
})

sample_transformed = preprocessor.transform(sample)
prediction = model.predict(sample_transformed)
probability = model.predict_proba(sample_transformed)

print("Prediction  :", "✅ Will Purchase" if prediction[0] == 1 else "❌ Will Not Purchase")
print("Probability :", f"Not Purchased = {probability[0][0]*100:.1f}%  |  Purchased = {probability[0][1]*100:.1f}%")

---

🚀 How to Run

1. Clone the repo

git clone https://github.com/AnmolPatel20/Holiday-Package-Prediction-Randomforest.git
cd Holiday-Package-Prediction-Randomforest

2. Install dependencies

pip install pandas numpy matplotlib seaborn scikit-learn joblib

3. Run the notebook

jupyter notebook Holiday_Package_Prediction.ipynb

---

📌 Notes

• Both holiday_package_classification_model.pkl and preprocessor.pkl must be loaded together for prediction
• The preprocessor handles all encoding and scaling — never pass raw data directly to the model

---

🙋 About

I'm on my machine learning journey — building, experimenting and documenting as I go. Every notebook here represents something I've genuinely tried to understand, not just run. 🚀

• GitHub: @AnmolPatel20
• Portfolio: anmolpatel20.github.io/Anmol_Portfolio

🙏 Acknowledgements

Thanks to Krish Naik Sir whose Udemy course has been a great resource throughout this learning journey.

"Not all those who wander are lost." — J.R.R. Tolkien

---

⭐ Star this repo if you found it helpful!