A web application built with Django, Auth-insight is intended to improve user authentication procedures for companies. Auth-insight seeks to address typical authentication issues including tiredness, dissatisfaction, and overload of logins by combining big data analysis and machine learning. The application includes data gathering for analysis, insights production based on machine learning categorization, and a single sign-on (SSO) mechanism for simplified authentication across various services. Ongoing authentication system optimization is ensured by constant monitoring. AuthInsight offers practical insights to enhance enterprise security and user experiences.
- Single Sign-On (SSO): Streamline user authentication across multiple organizational services with a seamless login experience.
- Big Data Integration: Collect, store, and analyze large volumes of user authentication data to gain valuable insights.
- Machine Learning Classification: Utilize machine learning algorithms to classify user activities and identify authentication-related challenges.
- Insights Generation: Generate actionable insights into common issues such as login overload, login fatigue, and login frustration.
- Continuous Monitoring: Deployed solution includes monitoring mechanisms for ongoing optimization and performance evaluation.
The clustering analysis results are shown and discussed in this section. Finding clusters and anomalies in user authentication activities is the goal of using KMeans clustering to the authentication data. To understand the patterns and trends in the data, the clustering findings are looked at. Nonetheless, the clustering approach acknowledges several limitations and constraints, such as inconclusive results and challenges in identifying meaningful clusters.
The integration of Random Forest into the authentication system provided significant improvements in the analysis of user authentication data. After extensive preprocessing and hyperparameter tuning, the Random Forest model demonstrated superior performance and accuracy. Below are the detailed results and analysis.
The Random Forest model's results underscore its effectiveness in handling the authentication data compared to the previously attempted KMeans clustering. The ensemble nature of Random Forest, combined with its capability to manage both categorical and numerical features, provided more reliable and accurate results. The model's precision, particularly in identifying successful authentication attempts, suggests that it can be a valuable tool in enhancing the security and user experience of the authentication system.
Future efforts will focus on addressing the lower performance metrics for unsuccessful authentication attempts and exploring additional features or alternative machine learning models to further optimize the system's accuracy and robustness.
- Clone the repository:
git clone https://github.com/Nathius262/auth-insight.git
cd auth-insightinstall dependencies
pip install -r requirements.txt
make migrations
python manage.py migrate
Run server
python manage.py runserver
Contributions are welcome! Please fork the repository and submit a pull request with your improvements.
This project is licensed under the MIT License.
Stephan Wiefling, Paul René Jørgensen, Sigurd Thunem, and Luigi Lo Iacono: Pump Up Password Security! Evaluating and Enhancing Risk-Based Authentication on a Real-World Large-Scale Online Service. In: ACM Transactions on Privacy and Security (2022). doi: 10.1145/3546069