exobengal 1.1.1

A python package to train Machine Learning models on Kepler Datasets with customizable hyperparameters to detect exoplanet

ExoBengal

Standardized tools for ML-based exoplanet candidate classification on NASA data, plus a companion docs website.

This repo contains:

• Python package exobengal:
  • DetectExoplanet for training/inference (RandomForest, CNN, kNN)
  • ExoParams convenience container for feature inputs
• Pretrained model artifacts in models/
• Example dataset(s) in data/
• Next.js docs site in website/

Full documentation is in docs/:

• Installation and requirements: docs/installation.md
• API reference: docs/api.md
• Data reference and preprocessing: docs/data.md
• Models and artifacts: docs/models.md
• Notebook walkthrough: docs/notebook.md

Quick Start

Install the package from PyPI (or your local environment):

pip install exobengal

Make a prediction with the bundled RandomForest model:

from exobengal.exobengal import DetectExoplanet

detector = DetectExoplanet()
sample = [365.0, 1.0, 288.0, 1.0, 4.44, 5778, 0.1, 5.0, 100.0]
print(detector.random_forest(sample))

Project Structure

exobengal/           # Python package
  exobengal.py      # DetectExoplanet class and helpers
models/              # Trained models (.pkl, .h5, scaler)
data/                # Dataset CSVs
website/             # Next.js static website + docs

Python API (quick view)

Constructor:

DetectExoplanet(
  rf_model_path="models/random_forest_classifier.pkl",
  cnn_model_path="models/cnn_model.h5",
  knn_model_path="models/knn_model.pkl",
  scaler_path="models/scaler.pkl",
  imputer_path="models/imputer.pkl",
)

Training:

detector.train_random_forest(data_path="data/cumulative_2025.09.20_12.15.37.csv")
detector.train_cnn(data_path="data/cumulative_2025.09.20_12.15.37.csv")
detector.train_knn(data_path="data/cumulative_2025.09.20_12.15.37.csv")

Inference (all return the same schema):

from exobengal.exobengal import ExoParams

sample = [koi_period, koi_prad, koi_teq, koi_srad, koi_slogg, koi_steff, koi_impact, koi_duration, koi_depth]
detector.random_forest(sample)
detector.cnn(sample)
detector.knn(sample)

# Or use ExoParams for clarity
params = ExoParams(period=365.0, prad=1.0, teq=288.0, srad=1.0, slog_g=4.44, steff=5778, impact=0.1, duration=5.0, depth=100.0)
detector.random_forest(params)

Utility:

detector.calculate_esi(koi_prad=1.05, koi_teq=290)

For full API and feature details, see docs/api.md.

Models

Artifacts live in models/. See docs/models.md for details and retraining notes.

Requirements

Python 3.8+. See docs/installation.md or requirements.txt.

Website

The Next.js site (in website/) includes a docs experience. See its README.md for running locally.

Development

Website (Node 20):

cd website
npm ci
npm run dev

Static export is enabled via output: 'export' and deployed to GitHub Pages with Actions.

License

MIT License – see LICENSE.