Welcome to the Web-page-Phishing-Detection project! This repository contains a complete workflow for detecting phishing web pages using state-of-the-art machine learning techniques. 🚦🔒

Phishing attacks are a major threat to online security. This project leverages machine learning to automatically classify web pages as phishing or legitimate based on a rich set of features extracted from URLs and page content.

Web-page-Phishing-Detection-main/
│
├── data/
│   └── dataset_phishing.csv         # Main dataset with 11,430 samples & 89 features
│
├── models/
│   ├── adaboost_classifier.pkl
│   ├── bagging_classifier.pkl
│   ├── decision_tree.pkl
│   ├── ensemble_voting_classifier_top5.pkl
│   └── ...                         # Other trained model files
│
├── best_model.pkl                  # Best performing model (Random Forest)
├── phishing_model_random_forest.pkl
├── phishing_model_random_forest_exp2.pkl
├── phishing_model.pkl
│
├── workflow_notebook.ipynb         # Main ML workflow notebook
├── workflow_notebook_experiment2.ipynb
├── Hyperparameter_tuning_workflow_notebook_experiment2.ipynb
│
├── README.md                       # This file!
├── .gitignore
├── LICENSE
│
├── CMP4040 - Project Document.docx # Project documentation
├── Team 9 - ML Project Report.docx
├── Team 9 - ML Project Report.pdf
└── ...

• Data Preprocessing:

  • Loads and cleans dataset_phishing.csv (see workflow_notebook.ipynb).
  • 89 features per sample, including URL structure, content analysis, and more.

• Model Training:

  • Multiple ML models implemented: Decision Tree, Random Forest, AdaBoost, Bagging, and Ensemble Voting.
  • Hyperparameter tuning for optimal performance (Hyperparameter_tuning_workflow_notebook_experiment2.ipynb).

• Evaluation & Selection:

  • Models evaluated for accuracy, precision, recall, and F1-score.
  • Best model saved as best_model.pkl.

• Experimentation:

  • Alternative workflows and experiments in workflow_notebook_experiment2.ipynb.

• Documentation:

  • Detailed project and team reports included for reference.

1. Clone the repository:

   git clone https://github.com/<your-username>/Web-page-Phishing-Detection-main.git

2. Explore the notebooks:

   • Open workflow_notebook.ipynb for the main workflow.
   • Use Hyperparameter_tuning_workflow_notebook_experiment2.ipynb for tuning experiments.

3. Run the models:

   • Load any .pkl model and use it to predict phishing URLs.

• workflow_notebook.ipynb
  Main pipeline: data loading, feature engineering, model training, evaluation.

• workflow_notebook_experiment2.ipynb
  Alternative experiments and model comparisons.

• Hyperparameter_tuning_workflow_notebook_experiment2.ipynb
  Hyperparameter optimization for improved accuracy.

• data/dataset_phishing.csv
  • 11,430 web page samples
  • 89 features per sample
  • Label: phishing or legitimate

• Random Forest (best model)
• Decision Tree
• AdaBoost
• Bagging
• Ensemble Voting

All models are saved in the models/ folder as .pkl files for easy reuse.

• Project Reports:
  • CMP4040 - Project Document.docx
  • Team 9 - ML Project Report.docx / .pdf

• Developed by Mamdouh Attia
• Licensed under the MIT License

Ready to fight phishing?
Dive into the notebooks, train your own models, or use the pre-trained ones to protect users from malicious web pages!

Stay safe. Stay smart. Detect phishing with ML!