D3d_100 dataset for d3d_loaders
This repository includes the definition of the D3D-100 dataset as well as routines to instantiate
it so that it works with D3D_loader.
Instantiating the dataset requires a copy of the d3d_signals repository to translate predictor aliases to MDS/PTDATA
point names. See the comments
in instantiate.sh
to see where to clone it.
Run
$ instantiate.sh
to instantiate the dataset.
This will fetch the data from D3D and calculate normalizations. Make sure to run this script on a machine that has access to the D3D MDSplus server.
Output of this script includes
- HDF5 data for the specified predictors in the subdirectory
D3D_100
. - normalization coefficients (mean and standard deviation) in the file
normalization.yaml