Cybersecurity Threat Classification using Machine Learning

Overview

This project classifies network intrusions using machine learning models. It utilizes the CICIDS2017 dataset to detect various attack types. The model used in this project is implemented in a single Jupyter Notebook file instead of separate scripts.

Requirements

Before running the code, ensure you have the following dependencies installed:

pip install pandas numpy scikit-learn matplotlib seaborn

Alternatively, you can install the dependencies using a requirements file:

pip install -r requirements.txt

Running the Code

1. Clone the Repository

   git clone https://github.com/2000pawan/Cyber-Attack-Threat-Classification.git
   

2. Download and Extract the Dataset

   • Download the CICIDS2017 dataset from here.
   • Extract the dataset into the data/ directory.

3. Run the Jupyter Notebook

   • Open the Jupyter Notebook environment:

   jupyter notebook
   

   • Navigate to the Cyber_attack.ipynb file and execute all cells sequentially.

Notebook Structure

• Data Preprocessing: Cleans missing values, normalizes numerical features, encodes categorical data, and balances the dataset using SMOTE.
• Model Training: Trains Random Forest and SVM classifiers with feature selection and hyperparameter tuning.
• Evaluation: Computes accuracy, precision, recall, and F1-score to compare model performance.
• Visualization: Generates confusion matrices and ROC curves for model assessment.

Results Summary

• Random Forest outperformed SVM in accuracy and generalization.
• Feature selection improved computational efficiency.
• could involve deep learning models for better performance.