Maxwell-Jia/kepler_flare

Kepler Flare Dataset

Dataset Description

The Kepler Flare Dataset is a comprehensive collection of stellar flare events observed by the Kepler Space Telescope. This dataset is constructed based on the flare event catalog presented in Yang & Liu (2019), which provides a systematic study of stellar flares in the Kepler mission. All light curves are the long cadence (about 29.4 minutes) data.

Dataset Features & Content

The dataset consists of high-quality photometric measurements from the Kepler Space Telescope, including:

• Light curves (flux measurements) normalized by their median values
• Binary classification labels for each time step (0: no flare, 1: flare event)
• Variable-length time series data
• Complete metadata for each observation

Dataset Structure

Each sample contains the following fields:

{
    'flux_norm': List[float],      # Normalized flux values sequence
    'label': List[bool],           # Corresponding labels (True for flare events)
    'metadata': {
        'kic_id': str,            # Kepler Input Catalog ID
        'quarter': str,           # Observation quarter
        'file_name': str,         # Original file name
        'flux_median': float,     # Median flux used for normalization
    }
}

Dataset Splits

The dataset is split into:

• Training set: 80%
• Validation set: 5%
• Test set: 15%

Usage

You can load the dataset directly using the Hugging Face datasets library:

from datasets import load_dataset

dataset = load_dataset("Maxwell-Jia/kepler-flare")

Applications

This dataset is particularly useful for:

• Stellar flare detection and analysis
• Astronomical event detection
• Deep learning applications in astronomical time series

Citation

If you use this dataset, please cite the following papers:

@article{yang2019flare,
  title={The flare catalog and the flare activity in the Kepler mission},
  author={Yang, Huiqin and Liu, Jifeng},
  journal={The Astrophysical Journal Supplement Series},
  volume={241},
  number={2},
  pages={29},
  year={2019},
  publisher={IOP Publishing}
}

@article{jia2024fcn4flare,
  title={FCN4Flare: Fully Convolution Neural Networks for Flare Detection},
  author={Jia, Ming-Hui and Luo, A-Li and Qiu, Bo},
  journal={arXiv preprint arXiv:2407.21240},
  year={2024}
}

License

This dataset is released under the MIT License.

Version

Current version: 1.0.0