alirezaaminzadeh/vuln-remediation-survival-model

Vulnerability Remediation Survival Model

Cox Proportional Hazards model for predicting time-to-remediation of enterprise vulnerabilities.

Model Type

• Algorithm: Cox Proportional Hazards (analytical coefficients)
• Target: Time from discovery to remediation (right-censored for open vulns)
• Concordance index: See train_metrics.json

Covariates

Feature	Effect
internet_exposed	Increases remediation hazard
patch_available	Increases remediation hazard
legacy_asset	Decreases remediation hazard
kev_status	Increases remediation hazard
cvss_score_norm	Increases with severity
epss_score	Increases with exploitation probability
asset_criticality	Increases with tier
environment_prod	Production prioritization

Files

• config.json — Model metadata
• cox_coefficients.json — Full Cox PH summary with hazard ratios
• survival_analysis.json — Pre-computed KM curves and comparisons
• train_metrics.json — Evaluation metrics

Intended Use

• Vulnerability management backlog prioritization
• SLA forecasting and risk tier assignment
• Security data science portfolio demonstration

Limitations

• Trained on synthetic data with known generative parameters
• Not validated on real scanner exports
• Does not replace human prioritization for zero-days

Team

Aria AI Security Data Science — aria-ai.ir