Based on the open datasets of ASAS-SN, Gaia and ZTF, we construct a compatible light curve dataset named LEAVES for automated recognition of variable stars, from the work " LEAVES: An Expandable Light Curve Dataset for Automatic Classification of Variable Stars ". The dataset contains a total of 979,962 variable and 134,592 non-variable light curves, in which the supported variables are divided into 6 superclasses and 9 subclasses. We validate the compatibility of the data set through experiments and use it to train a hierarchical random forest classifier.
We provide the light curves of these objects by class in csv files in https://drive.google.com/drive/folders/1R8r4oBXKxLdC6puUkS4IUx4VTY_lTluW. The composition of LEAVES is shown in table below:
| Class | Subclass | ASAS-SN_V | ASAS-SN_g | ZTF | Gaia | Total |
|---|---|---|---|---|---|---|
| Eclipsing binaries(EB) | EA | 37463 | 10695 | 29198 | 6785 | 421025 |
| EW | 57058 | 22724 | 257093 | |||
| Rotational variables(ROT) | ROT | 16850 | 4202 | 98060 | 119112 | |
| RR Lyrae(RR) | RRAB | 13244 | 10981 | 17598 | 840 | 61366 |
| RRC | 5959 | 3624 | 9120 | |||
| Cepheids(CEP) | CEP | 1003 | 405 | 1074 | 2 | 2484 |
| Long period variables(LPV) | SR | 140997 | 143376 | 46738 | 5095 | 355374 |
| M | 7884 | 6133 | 5151 | |||
| Delta Scuti(DSCT) | DSCT | 3718 | 939 | 13944 | 18601 | |
| Non-var | Non-var | 134592 | 134592 |
The full data set and machine learning models are also available and will be released officially at the China National Astronomical Data Center.