ИСТИНА |
Войти в систему Регистрация |
|
ИПМех РАН |
||
Machine learning approaches and their subsets such as neural networks and deep learning methods are widely applied to improve traditional space weather forecasts based on operational physical models which use data from various satellites. SatData has been designed as a tool that helps data scientists to avoid the challenging data acquisi- tion step and focus on data analysis. Currently, SatData supports fetching data from Russian satellites (Lomonosov, Meteor-M1, Meteor-M2, Electro-L1, Electro-L2, etc.), American satellites (ACE, GOES series, DSCOVR, SDO, etc.), space indices (Kp, Dst, Ap, .etc.) as well as ground stations. SatData represents satellite data as Python arrays that can be passed directly to scikit-learn classes.