satellite-anomaly-detection

Kaggle release assets and reproducibility code for:

Hybrid Satellite Telemetry Anomaly Detection: A Novel Dataset, Fault Taxonomy, Recurrence-Plot Computer Vision, and Comparative Machine Learning Study

• Paper page: https://aryanputta.com/papers/hybrid-satellite-telemetry-anomaly-detection.html
• PDF: https://aryanputta.com/assets/papers/hybrid-satellite-telemetry-anomaly-detection.pdf
• Portfolio: https://aryanputta.com/

Project Structure

satellite-anomaly-detection/
├── data/
│   └── generate_dataset.py
├── encoding/
│   └── recurrence_plot.py
├── evaluation/
│   └── fault_type_eval.py
├── demo/
│   └── live_demo.py
├── noise/
│   └── arduino_noise_capture/
│       ├── noise_capture.ino
│       └── noise_log.csv
├── models/
├── notebooks/
│   └── satellite_anomaly_detection.ipynb
├── kaggle/
│   ├── dataset_page.md
│   ├── benchmark_notebook.ipynb
│   └── research_report.md
├── scripts/
│   ├── sync_blank_app.py
│   └── validate_kaggle_artifacts.py
├── requirements.txt
└── README.md

This repo now keeps only the two Kaggle upload artifacts:

• kaggle/dataset_page.md (dataset page content)
• kaggle/benchmark_notebook.ipynb (notebook upload file)

Owner/account name: aryanputta

• Dataset: https://kaggle.com/datasets/aryanputta/hybrid-satellite-telemetry
• Code: https://github.com/aryanputta/satellite-anomaly-detection

How to import your own code/files here

Option A: Copy local files directly

From your machine, copy your files into matching folders above.

Option B: Use git add + commit

1. Put files in the right folders (for example models/, notebooks/, noise/arduino_noise_capture/).
2. Run:

git add .
git commit -m "Add my models and experiment files"

Option C: Replace notebook with your own

If you want your exact notebook:

1. Save your notebook as satellite_anomaly_detection.ipynb
2. Place it at notebooks/satellite_anomaly_detection.ipynb
3. Run validation:

python scripts/validate_kaggle_artifacts.py
python -m json.tool notebooks/satellite_anomaly_detection.ipynb > /tmp/notebook_ok.json

Option D: Paste your .py files directly

If you paste code in chat, I can place it directly into the correct file path for you. For example:

• evaluation/fault_type_eval.py
• data/generate_dataset.py
• demo/live_demo.py
• encoding/recurrence_plot.py

I already used this flow for the latest update.

Quick start

Install dependencies:

pip install -r requirements.txt

Generate dataset:

python data/generate_dataset.py --output hybrid_satellite_dataset

Run validator:

python scripts/validate_kaggle_artifacts.py

Check public links:

python scripts/check_links.py

Paper-coverage audit:

python scripts/audit_paper_coverage.py

Run sequence after you add your CSV

1. Verify the dataset file is present and has expected columns/size:

python scripts/check_data_ready.py --csv data/hybrid_satellite_telemetry.csv --min-rows 100000

2. Run model evaluation (all models):

python evaluation/fault_type_eval.py --all-models --data data/hybrid_satellite_telemetry.csv

3. Run Kaggle artifact validation:

python scripts/validate_kaggle_artifacts.py

If Python dependencies are missing in your environment, install with:

pip install -r requirements.txt

Kaggle release assets for:

Hybrid Satellite Telemetry Anomaly Detection: A Novel Dataset, Fault Taxonomy, Recurrence-Plot Computer Vision, and Comparative Machine Learning Study

• Dataset: https://kaggle.com/datasets/aryantputta/hybrid-satellite-telemetry
• Code: https://github.com/aryanputta/satellite-anomaly-detection
• kaggle/dataset_page.md: Kaggle dataset page draft
• kaggle/benchmark_notebook.ipynb: Kaggle benchmark notebook draft