Project detail
Security Risks with PyMC
Bayesian model for estimating breach probability and detecting suspicious events in network patterns.
Highlights
- Probabilistic detection of malicious events
- Risk analysis with network variables
- Configurable threshold for high risk
- Foundation for prioritizing defenses
No public repository or demo link is published for this project yet.
Architecture
Applied guide with Bayesian logistic regression on traffic patterns and high-risk observation classification
Challenges
- Capturing noisy signals without losing detection capability
- Reducing false positives and false negatives
- Explaining risk usefully for operations
Outcome
A more interpretable approach to prioritizing security events and evaluating risk with explicit uncertainty
Overview
Problem
Security needs to react to changing signals without relying solely on rigid rules.
Approach
I modeled malicious event probability with PyMC using traffic variables and network context.
Outcome
The model helps classify suspicious events and focus attention where risk is highest.