Back to Projects
Completed Project

Cross-Dataset Threat Detection Benchmark

A reproducible security-ML laboratory that compares anomaly classifiers across multiple traffic datasets, exposing model drift, data leakage, class imbalance, calibration errors, and cross-environment performance loss.

|
Cross-Dataset Threat Detection Benchmark
RoleExperimental design and ML evaluation
TimelineData pipeline • Benchmark design
TeamIndependent project
Tech Stack5 Technologies

Mission Brief

The benchmark standardizes ingestion, preprocessing, semantic feature mapping, feature selection, class-imbalance handling, model training, probability calibration, and experiment tracking. Logistic Regression, Decision Tree, SVM, Random Forest, and gradient-boosted models are evaluated within their source dataset and against an independent traffic distribution. Model cards report performance by attack category together with inference latency, model size, calibration quality, and explainability evidence.

Engineering Chronicles

Different datasets use incompatible schemas and traffic distributions.

SolutionCreate a shared semantic feature layer and report unsupported features explicitly rather than silently imputing meaning.

Project Access

Technologies

Python
Pandas
scikit-learn
XGBoost
SHAP

Table of Contents

  • • Mission Brief
  • • Engineering Chronicles
  • • Visual Gallery