1. Abstract

This project focuses on detecting phishing websites using machine learning techniques. Phishing websites are malicious websites that attempt to steal sensitive information such as login credentials, credit card details, and personal data by pretending to be trustworthy entities.

In this project, a dataset containing metadata information of 5000 legitimate websites and 5000 phishing websites is used. The dataset includes various features that describe the characteristics of these websites. Understanding the meaning of these metadata features is important before building the machine learning model.

Data preprocessing and feature selection techniques are applied to identify the most relevant features that help distinguish phishing websites from legitimate ones. A Random Forest Classifier machine learning model is trained using the processed dataset to classify whether a website is phishing or legitimate.

After building the prediction model, a web application is developed using the Django framework. The trained machine learning model is integrated into this application so users can check whether a given website is phishing or safe.

Finally, the Django web application is deployed on the Heroku cloud platform using GitHub integration, making the system accessible online. This project demonstrates how machine learning and web technologies can be used to improve cybersecurity by detecting phishing websites.

2. Objectives

The main objectives of this project are:

1. To understand phishing attacks and their impact on cybersecurity.
2. To analyze metadata information related to websites.
3. To preprocess and analyze the phishing dataset.
4. To perform feature selection to identify important features.
5. To build a classification model using the Random Forest Classifier algorithm.
6. To train and evaluate the model for phishing website detection.
7. To develop a web application using the Django framework.
8. To deploy the application on Heroku using GitHub integration.

3. Existing System

In the existing system, phishing detection is mainly performed using traditional security tools, manual verification, or rule-based detection systems.

This system has several limitations:

1. Manual detection of phishing websites is time-consuming.
2. Traditional systems rely on predefined rules that may not detect new phishing techniques.
3. Large volumes of websites make manual analysis difficult.
4. Some phishing websites closely resemble legitimate websites, making detection challenging.
5. Existing systems may not automatically learn from new data patterns.

Due to these limitations, detecting phishing websites efficiently becomes difficult using traditional approaches.

4. Proposed System

The proposed system uses machine learning techniques to automatically detect phishing websites.

In this system:

1. A dataset containing metadata of legitimate and phishing websites is used.
2. Data preprocessing is performed to clean and prepare the dataset.
3. Feature selection techniques are applied to identify the most important features.
4. A Random Forest Classifier model is trained to classify websites as phishing or legitimate.
5. A Django-based web application is developed to provide an interactive interface.
6. Users can input website information to check whether the site is phishing or safe.
7. The application is deployed online using Heroku and GitHub integration.

This system provides faster and more accurate phishing detection compared to manual methods.

5. Implementation Procedure

The implementation of this project is carried out in the following steps:

Step 1: Data Collection

1. Obtain a dataset containing metadata information of phishing and legitimate websites.

Step 2: Data Preprocessing

1. Clean the dataset and remove inconsistencies.
2. Understand the meaning of the metadata features.

Step 3: Feature Selection

1. Analyze features to determine which ones are most important for detecting phishing websites.

Step 4: Model Development

1. Select the Random Forest Classifier algorithm.
2. Train the machine learning model using the phishing dataset.

Step 5: Model Evaluation

1. Test the model using test data.
2. Evaluate the model performance using classification metrics such as accuracy.

Step 6: Web Application Development

1. Develop a web application using the Django framework.
2. Integrate the trained machine learning model into the Django application.

Step 7: Deployment

1. Upload the project to GitHub.
2. Deploy the Django web application on Heroku cloud platform.
3. Make the application accessible online.

6. Software Requirements

The software used in this project includes:

1. Operating System: Windows / Linux / macOS
2. Programming Language: Python 3.x
3. Framework: Django
4. IDE: Jupyter Notebook / VS Code / PyCharm

Libraries:

1. NumPy
2. Pandas
3. Matplotlib
4. Seaborn
5. Scikit-learn

Deployment Tools:

1. GitHub
2. Heroku

Web Browser: Chrome / Firefox

7. Hardware Requirements

The hardware required for this project includes:

1. Processor: Intel i3 / i5 or higher
2. RAM: Minimum 4 GB (8 GB recommended)
3. Storage: Minimum 128 GB free space
4. System: Laptop / Desktop Computer
5. Internet Connection

8. Advantages of the Project

1. Automatically detects phishing websites using machine learning.
2. Improves cybersecurity by preventing phishing attacks.
3. Handles large datasets efficiently.
4. Uses feature selection for better model performance.
5. Provides a user-friendly web interface.
6. Allows users to check websites in real time.
7. Easily deployable on cloud platforms.
8. Demonstrates integration of machine learning with web development.