Python wrappers for the Matlab code for the Import Vector Machine (IVM) classifier from Freie University (http://www.ipb.uni-bonn.de/ivm/?L=1)

This assumes the following:

• you already have installed mlabwrap (http://mlabwrap.sourceforge.net/)
• The Matlab executable is on your path
• You have configured mex to work correctly

Installation Steps:

• Clone the github or download the source code
• Download the ivmSoftware4.3 from (http://www.geo.fu-berlin.de/en/geog/fachrichtungen/geoinformatik/medien/download/ivmSoftware4_3.zip)
• Extract the zip file to ivm-wrapper/ivm/ivmSoftware4.3
• run 'python setup.py install'

The wrapper extends scikit-learn base classifier so it should be compatible with all the scikit-learn extras such as grid_search and cross_val_score. A simple test on the iris dataset:

from sklearn import datasets
from sklearn.cross_validation import cross_val_score
from ivm import IVM

data = datasets.load_iris()
clf = IVM()
print cross_val_score(clf, data.data, data.target)