This content refers to the previous stable release of PyMVPA. Please visit www.pymvpa.org for the most recent version of PyMVPA and its documentation.

Visualization of Data Projection Methods

from mvpa.misc.data_generators import noisy_2d_fx
from mvpa.mappers.pca import PCAMapper
from mvpa.mappers.svd import SVDMapper
from mvpa.mappers.ica import ICAMapper
from mvpa import cfg

import pylab as P
import numpy as N
center = [10, 20]
axis_range = 7

def plotProjDir(mproj):
    p = mproj + N.array(center).T

    P.plot([center[0], p[0,0]], [center[1], p[0,1]], hold=True)
    P.plot([center[0], p[1,0]], [center[1], p[1,1]], hold=True)



mappers = {
            'PCA': PCAMapper(),
            'SVD': SVDMapper(),
            'ICA': ICAMapper(),
          }
datasets = [
    noisy_2d_fx(100, lambda x: x, [lambda x: x],
                center, noise_std=.5),
    noisy_2d_fx(50, lambda x: x, [lambda x: x, lambda x: -x],
                center, noise_std=.5),
    noisy_2d_fx(50, lambda x: x, [lambda x: x, lambda x: 0],
                center, noise_std=.5),
   ]

ndatasets = len(datasets)
nmappers = len(mappers.keys())

P.figure(figsize=(8,8))
fig = 1

for ds in datasets:
    for mname, mapper in mappers.iteritems():
        mapper.train(ds)

        dproj = mapper.forward(ds.samples)
        mproj = mapper.proj
        print mproj

        P.subplot(ndatasets, nmappers, fig)
        if fig <= 3:
            P.title(mname)
        P.axis('equal')

        P.scatter(ds.samples[:, 0],
                  ds.samples[:, 1],
                  s=30, c=(ds.labels) * 200)
        plotProjDir(mproj)
        fig += 1


if cfg.getboolean('examples', 'interactive', True):
    P.show()

Output of the example:

See also

The full source code of this example is included in the PyMVPA source distribution (doc/examples/projections.py).