Astraea

Astraea is a package to train Random Forest (RF) models on datasets. It provides tools to train RF classifiers and regressors as well as perform simple cross-validation tests and create performance plots for the test set.

It was first developed to calculate rotation period of stars from various stellar properties provided and is intended to predict long rotation periods (e.g. those of M-dwarfs) from short TESS lightcurves (27-day lightcurves).

We provide access to models trained on stars from the catalog by McQuillan et al. (2014), Garcia et al. (2014), and Santos et al. (2019). Users can predict whether rotation period can be recovered and predict recoverable rotation periods for the stars in the Kepler field by using their temperatures, colors, kinematics, and other stellar parameters.

User Guide

• Installation
• Dependencies
• Running tests
• API documentation

Tutorials

• Predict rotation period for Kepler stars using existing model
  • Calculate rotation period(s) from features
  • Download a light curve and calculate variablity, lomb-scargle peak and flickers needed for the trained model
• Train a regressor model and test its performance
  • Train a regressor model
  • Plot the feature importances and Predicted vs True plot
  • Use the trained model to predict a new label value
• Train a classifier model and plot the Receiver operating characteristic (ROC) curve
  • Train a classification model
  • Plot the ROC curve

License & attribution

Copyright 2020, Yuxi Lu.

The source code is made available under the terms of the MIT license.

If you make use of this code, please cite this package and its dependencies. You can find more information about how and what to cite in the citation documentation (not yet complete).