Data-driven stellar intrinsic colors and dust reddenings for spectro-photometric data

Intrinsic colors of stars are essential for the studies on both stellar physics and dust reddening. In this work, we developed an XGBoost model to predict the stellar intrinsic colors with the atmospheric parameters, Teff , log g, and [M/H], which is an improvement of the widely used blue-edge method. The dust reddening toward each line-of-sight can then be calculated by the observed colors minus the derived intrinsic colors. Here we provide the related data sets: xgb_model.pkl: the trained XGBoost model. use_xgb.py: a simple script showing how to use the XGBoost model to predict intrinsic colors. data_set.fits: this fits file contains the training and test sets, separating into four data arrays: X_train: X-data (teff,logg,mh) of the training set. X_test: X-data (teff,logg,mh) of the test set. y_train: y-data (BP-RP, BP-Ks, J-Ks) of the training set. y_test: y-data (BP-RP, BP-Ks, J-Ks) of the test set. With above data and model, users can apply the trained XGBoost to new sources to predict their intrinsic colors and calculate their dust reddenings. One can further control the quality of the prediction by selecting training-like sources with the training set and estimating the generalization error by the test set.