PLAsTiCC and the Kaggle Dataset

The dataset comes from the Kaggle competition "Photometric LSST Astronomical Time Series Classification Challenge" (PLAsTiCC) [1]. Held in late 2018, the competition was in preparation for first light of the Large Synoptic Survey Telescope (LSST) [2].

LSST endeavors to detect transients - stars that are actively changing - and notify astronomers of their presence. Examples of transients are:

  • a Supernova that explodes over a 100 day period
  • a Pulsar that flashes once every 12 hours
  • a Lensing Event (a planet goes in front of a star) that occurs... occasionally!
  • Each night LSST will capture 20 Terabytes of data and and classify about 2.5 billion stars and galaxies. The goal is:

  • 60 minutes (image capture to notification)
  • It is a huge classification challenge!

    ???

    References:

  • (1) https://arxiv.org/abs/1810.00001
  • (2) https://www.lsst.org/