If you are just hacking around, please put your code in scripts -- it's for any one-off analysis or snippet of code that isn't yet ready for reuse by other people. We recommend using scripts/demo_script.py as a guide for the types of operations that you may find useful.
There is an xforms system for saving complicated analyses in a way that you can reload later; this was what we used in the scripts/fit_model.py command. We also have a good tutorial/template at scripts/demo_script.py. We recommend beginners make a copy of it and edit it as needed. You may run it with:
# Run the demo script
python3 scripts/demo_script.py
The tests/ directory is intended for unit tests that run automatically with pytest.
If your data is in Recording format, put it in the recordings/ directory. If it is not yet in recording form, you may want to make a conversion script in scripts/ that is able convert your data from your custom format into Signals and Recordings.
Below is some pseudocode for a script that converts your data from a custom format and then save it as a Signal to make it easier for other people to use.
from nems.signal import Signal
numpy_array = load_my_custom_data_format(...)
sig = Signal(matrix=numpy_array,
name='mysignal',
recording='some-string',
fs=200 # Hz
)
sig.save('../recordings/my-new-recording/my-new-signal')
Once you have fit a model to a recording, you can save the resulting files either locally in your nems/results directory, or to a remote NEMS_DB server.