The Healing Analytics Engine

An end-to-end anomaly detection pipeline built inside Google's Antigravity IDE. It simulates a messy network traffic dataset, cleans it using pandas, trains an unsupervised machine learning model to detect extreme traffic outliers, and exports reports and visualizations.

---

📋 Pipeline Workflow

The pipeline performs four key operations sequentially:

graph TD
    A[1. Simulate Raw Data] -->|traffic_raw.csv| B[2. Data Cleaning]
    B -->|Cleaned DataFrame| C[3. Train Isolation Forest]
    C -->|Anomaly Labels| D[4. Generate Report & Plot]

Loading

1. Dataset Simulation:

   • Generates 10,000 rows of traffic data starting from 2026-06-01.
   • Injects messy timestamp formats (e.g. DD/MM/YYYY, prefix ERR-, suffix (UTC), and corrupted values).
   • Injects 5% missing values (NaNs) across all columns.
   • Injects extreme outliers (high traffic volume spikes, negative traffic volume, and high response latencies).

2. Data Cleaning & Imputation:

   • Standardizes timestamp strings by stripping prefixes/suffixes and parses them.
   • Drops rows with missing or corrupted timestamps.
   • Imputes missing numerical values (traffic_volume and response_time) using the median, which is robust to the injected extreme outliers.

3. Anomaly Classification:

   • Trains an Isolation Forest model (contamination=0.05) to isolate the top 5% extreme multivariate anomalies.

4. Reporting & Visualizations:

   • Exports anomaly_report.md summarizing cleaning stats and anomaly thresholds.
   • Generates anomaly_clusters.png, a scatter plot highlighting the anomalies in red against normal indigo data.

---

🛠️ Repository Structure

├── .github/workflows/
│   └── pipeline.yml       # GitHub Actions workflow
├── anomaly_clusters.png   # Matplotlib scatter plot (output)
├── anomaly_pipeline.py    # Main pipeline python script
├── anomaly_report.md      # Markdown report summary (output)
├── pyproject.toml         # Python project configuration & dependencies
├── README.md              # Documentation
├── traffic_raw.csv        # Simulated raw dataset
└── uv.lock                # Locked dependencies file

---

🚀 Getting Started

This project uses uv for fast, reliable dependency and environment management.

Prerequisites

Install uv if you haven't already:

# On macOS/Linux
curl -LsSf https://astral-sh/uv/install.sh | sh

Local Execution

Clone the repository and run the pipeline:

# Sync dependencies and run the script
uv run python anomaly_pipeline.py

This will automatically:

1. Detect or simulate traffic_raw.csv.
2. Run the cleaning, modeling, and plotting routines.
3. Overwrite/update anomaly_report.md and anomaly_clusters.png in the directory.

---

🤖 GitHub Actions Workflow

The repository includes a GitHub Actions configuration in .github/workflows/pipeline.yml. On every push or pull request to the main branch, the runner will:

• Check out the repository.
• Set up Python and uv.
• Execute the pipeline script (anomaly_pipeline.py).
• Upload the generated anomaly_report.md and anomaly_clusters.png as workflow artifacts.