Skip to content

Getting started

LCDC: Light curve dataset creator is a Python package that allows you to work with large light curve datasets in a simple and efficient way. It is designed to be used creation of dataset for machine learning models as it produces output in datasets.Dataset format. Also it is a powerful tool for data preprocessing and scientific analysis on whole populations.

It is sutaible for working with MMT_snapshot dataset created from MMT database 1.

Instalation

pip install lcdc

Simple Example

from lcdc import DatasetBuilder
from lcdc import vars
from lcdc import utils
import lcdc.preprocessing as pp
import lcdc.stats as stats

db = DatasetBuilder(DATA_PATH, norad_ids=[IDX])
preprocessing = [
    pp.FilterByPeriodicity(vars.Variability.PERIODIC),
    pp.SplitByRotationalPeriod(1), 
    pp.FilterMinLength(100),
    pp.FilterFolded(100, 0.8), 
]

db.preprocess(preprocessing)
dataset = db.build_dataset()
print(dataset)

Loaded 402 track
Preprocessing: 100%|██████████| 402/402 [00:08<00:00, 49.28it/s]
Dataset({
    features: ['norad_id', 'id', 'period', 'timestamp', 'time', 'mag', 'phase', 'distance', 'filter', 'name', 'variability', 'label', 'range'],
    num_rows: 4057
})

📝 Citing

@article{kyselica2025lcdc,
  title={LCDC: Bridging Science and Machine Learning for Light Curve Analysis},
  author={Kyselica, Daniel and Hrob{\'a}r, Tom{\'a}{\v{s}} and {\v{S}}ilha, Ji{\v{r}}{\'\i} and {\v{D}}urikovi{\v{c}}, Roman and {\v{S}}uppa, Marek},
  journal={arXiv preprint arXiv:2504.10550},
  year={2025}
}

  1. Karpov, S., et al. "Mini-Mega-TORTORA wide-field monitoring system with sub-second temporal resolution: first year of operation." Revista Mexicana de Astronomía y Astrofísica 48 (2016): 91-96. ↩