Frequently Asked Questions

I am tired of writing these endless import blocks. Any alternative?

Sure. Instead of individually importing all pieces that are required by a script, you can import them all at once. A simple:

import mvpa.suite as mvpa

makes everything directly accessible through the mvpa namespace, e.g. mvpa.datasets.base.Dataset becomes mvpa.Dataset. Really lazy people can even do:

from mvpa.suite import *

However, as always there is a price to pay for this convenience. In contrast to the individual imports there is some intial performance and memory cost. In the worst case you’ll get all external dependencies loaded (e.g. a full R session), just because you have them installed. Therefore, it might be better to limit this use to case where individual key presses matter and use individual imports for production scripts.

I feel like I want to contribute something, do you mind?

Not at all! If you think there is something that is not well explained in the documentation, send us an improvement. If you implemented a new algorithm using PyMVPA that you want to share, please share. If you have an idea for some other improvement (e.g. speed, functionality), but you have no time/cannot/do not want to implement it yourself, please post your idea to the PyMVPA mailing list.

The manual is quite insufficient. When will you improve it?

Writing a manual can be a tricky task if you already know the details and have to imagine what might be the most interesting information for someone who is just starting. If you feel that something is missing which has cost you some time to figure out, please drop us a note and we will add it as soon as possible. If you have developed some code snippets to demonstrate some feature or non-trivial behaviour, please consider sharing this snippet with us and we will put it into the example collection or the manual. Thanks!