Tutorial Introduction to PyMVPA¶
In this tutorial we are going to take a look at all major parts of PyMVPA, introduce the most important concepts, and explore particular functionality in real-life analysis examples. This tutorial also serves as basic course material for workshops on introductions to MVPA. Please contact us, if you are interested in hosting a PyMVPA workshop at your institution.
Please note that this tutorial is only concerned with aspects directly related to PyMVPA. It does not teach basic Python programming. If you are new to Python, we recommend that you take a look at the Tutorial Prerequisites for information about what you should know and how to obtain that knowledge.
Throughout the tutorial there will be little exercises with tasks that aim to deepen your understanding of a particular problem or to train important skills. However, even without a dedicated exercise you are advised to run the tutorial code interactively and explore code snippets beyond what is touched by the tutorial. Typically, only the most important aspects will be mentioned and each building block in PyMVPA can be used in more flexible ways than what is shown. Enjoy the ride.
Throughout the tutorial we will analyze real BOLD fMRI data. Therefore, to be able to run the code in this
tutorial, you need to download the corresponding data from the PyMVPA
website. Once downloaded, extract the tarball. On a
NeuroDebian-enabled system, the tutorial data is also available from the
pymvpa2-tutorial command (installed with PyMVPA) can be invoked in a
console in order to launch a tutorial session. If the tutorial data was
downloaded manually it may be necessary to specify the appropriate
--tutorial-data-path option (see
pymvpa2-tutorial --help for more
Virtually every Python script starts with some
import statements that load
functionality provided elsewhere. Likewise a tutorial session needs to import
the PyMVPA packages and some little helpers we are going to use in the
>>> from mvpa2.tutorial_suite import *
If this command succeeds without error, everything is ready to go.
If you want to prevent yourself from re-typing all code snippets into the
terminal window, you might want to investigate IPython’s
or use the `IPython notebooks`_ provided for each tutorial part.
- Tutorial Prerequisites
- Dataset basics and concepts
- Getting data in shape
- Classifiers – All Alike, Yet Different
- Looking here and there – Searchlights
- Classifiers that do more – Meta Classifiers
- Classification Model Parameters – Sensitivity Analysis
- Event-related Data Analysis
- Multi-dimensional Searchlights
- Working with OpenFMRI.org data
- WiP: The Earth Is Round – Significance Testing