# Intrusion_Dection_System

**Repository Path**: HeJiaxing97/Intrusion_Dection_System

## Basic Information

- **Project Name**: Intrusion_Dection_System
- **Description**: Build Intrusion Detection System via basic machine learning models and compare performance. 
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 2
- **Forks**: 0
- **Created**: 2020-07-15
- **Last Updated**: 2023-04-06

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Intrusion_Dection_System


This project aims to build an intrusion detection systems(IDS) and select the best binary classification model to determine whether an access might be malicious. We found a dataset containing five million server connection records in a simulated network environment from KDD. In terms of target variable, the original dataset contains a feature categorizing connections into either normal or 24 types of attack, and we plan to create a binary variable indicating if the connection is an attack as the target variable. Given that original KDD dataset is generated in 1999, there have been some analysis and paper published based on this dataset. After reading through some of the paper for reference and conducting our own baseline analysis, we decided to proceed forward with a certain type of machine learning models, including logistic regression, random forest, knn, decision tree and neural network. For each model, we conducted cross validation method as well as tuning certain hyperparameters to improve every model’s predictive ability and reduce common issues such as overfitting. We ultimately found out that our baseline model random forest is able to achieve the highest AUC score.