Context Solar flares are intense bursts of radiation which can disrupt the power grids of a continent, shut down the GPS system or irradiate people exposed in space. Developing systems for predicting solar flares would allow us to precisely aim our observation instruments at upcoming events, and eventually enable countermeasures against such worst-case scenarios. Content The SDOBenchmark dataset has a dedicated webpage at [i4ds.github.io/SDOBenchmark][1], where you will find plenty of information. And if things are still unclear, please don't hesitate to ask questions in the Discussion tab! Acknowledgements This dataset was created by Roman Bolzern and Michael Aerni from the [Institute for Data Science, FHNW, Switzerland][2]. We owe our thanks to [the SDO satellite mission][3], and to [JSOC Stanford][4] for providing the raw data. Inspiration The prediction of solar flares proves to be a challenging problem, some even compare it to weather forecasting. And regarding Machine Learning, we find this dataset to be particularly challenging because of the complexity of a single sample (up to 40 images), the relatively small size of samples (8'000 for training), and the fact that it is a regression problem. Yet Kagglers have proven time and time again that predictions can be made on the most complex of data. By providing this dataset, we hope to encourage Kaggle machine learners to push the envelope of solar flare predictions. [1]: https://i4ds.github.io/SDOBenchmark [2]: https://www.fhnw.ch/en/about-fhnw/schools/school-of-engineering/institutes/institute-for-data-science [3]: https://sdo.gsfc.nasa.gov/ [4]: http://jsoc.stanford.edu/