🩺 Breast Cancer Anomaly Detection

A lightweight unsupervised anomaly detection project using the Multivariate Gaussian Model to identify abnormal breast cancer cases. The model learns the normal data distribution and flags low-probability observations as potential anomalies.

---

🔍 Overview

• Applies density estimation to detect rare malignant patterns
• Uses multivariate Gaussian parameters (mean + covariance)
• Threshold (ε) selected by maximizing F1-score
• Visualizes outliers that deviate from the main cluster

This approach demonstrates how simple unsupervised methods can reveal meaningful medical anomalies without relying on labels.

---

📊 Dataset

• Wisconsin Breast Cancer Dataset (WBC)
• 569 samples, 30 standardized numeric features
• Labels used only for evaluation (not for training)

---

🧠 Method Highlights

• Feature preprocessing & normalization
• Gaussian probability density calculation
• Automatic threshold tuning
• Outlier detection & visualization

---

🛠️ Technologies

Python, NumPy, Pandas, Matplotlib, Seaborn