Obtaining SCoT

##### From PyPi

Use the following command to install SCoT from PyPi:

pip install scot

##### From Source

Use the following command to fetch the sources:

git clone –recursive https://github.com/scot-dev/scot.git scot

The flag –recursive tells git to check out the numpydoc submodule, which is required for building the documentation.

Documentation

Documentation is available online at http://scot-dev.github.io/scot-doc/index.html.

Dependencies

Required: numpy, scipy

Optional: matplotlib, scikit-learn

The lowest supported versions of these libraries are numpy 1.8.0, scipy 0.13.3, scikit-learn 0.15.0, and matplotlib 1.4.0. Lower versions may work but are not tested.

Examples

To run the examples on Linux, invoke the following commands inside the SCoT main directory:

PYTHONPATH=. python examples/misc/connectivity.py

PYTHONPATH=. python examples/misc/timefrequency.py

etc.

Note that you need to obtain the example data from https://github.com/SCoT-dev/scot-data. The scot-data package must be on Python’s search path.

Note

As of version 0.2, the data format in all SCoT routines has changed. It is now consistent with Scipy and MNE-Python. Specifically, epoched input data is now arranged in three-dimensional arrays of shape (epochs, channels, samples). In addition, continuous data is now arranged in two-dimensional arrays of shape (channels, samples).