This file is indexed.

/usr/share/pyshared/mvpa/clfs/knn.py is in python-mvpa 0.4.8-1.

This file is owned by root:root, with mode 0o644.

The actual contents of the file can be viewed below.

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
# emacs: -*- mode: python; py-indent-offset: 4; indent-tabs-mode: nil -*-
# vi: set ft=python sts=4 ts=4 sw=4 et:
### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ##
#
#   See COPYING file distributed along with the PyMVPA package for the
#   copyright and license terms.
#
### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ##
"""k-Nearest-Neighbour classifier."""

__docformat__ = 'restructuredtext'

import sys
# not worthy of externals checking
_dict_has_key = sys.version_info >= (2, 5)

import numpy as N

from mvpa.base import warning
from mvpa.misc.support import indentDoc
from mvpa.clfs.base import Classifier
from mvpa.base.dochelpers import enhancedDocString
from mvpa.clfs.distance import squared_euclidean_distance

if __debug__:
    from mvpa.base import debug


class kNN(Classifier):
    """
    k-Nearest-Neighbour classifier.

    This is a simple classifier that bases its decision on the distances
    between the training dataset samples and the test sample(s). Distances
    are computed using a customizable distance function. A certain number
    (`k`)of nearest neighbors is selected based on the smallest distances
    and the labels of this neighboring samples are fed into a voting
    function to determine the labels of the test sample.

    Training a kNN classifier is extremely quick, as no actuall training
    is performed as the training dataset is simply stored in the
    classifier. All computations are done during classifier prediction.

    .. note::
      If enabled, kNN stores the votes per class in the 'values' state after
      calling predict().

    """

    _clf_internals = ['knn', 'non-linear', 'binary', 'multiclass',
                      'notrain2predict' ]

    def __init__(self, k=2, dfx=squared_euclidean_distance,
                 voting='weighted', **kwargs):
        """
        :Parameters:
          k: unsigned integer
            Number of nearest neighbours to be used for voting.
          dfx: functor
            Function to compute the distances between training and test samples.
            Default: squared euclidean distance
          voting: str
            Voting method used to derive predictions from the nearest neighbors.
            Possible values are 'majority' (simple majority of classes
            determines vote) and 'weighted' (votes are weighted according to the
            relative frequencies of each class in the training data).
          **kwargs:
            Additonal arguments are passed to the base class.
        """

        # init base class first
        Classifier.__init__(self, **kwargs)

        self.__k = k
        self.__dfx = dfx
        self.__voting = voting
        self.__data = None


    def __repr__(self, prefixes=[]):
        """Representation of the object
        """
        return super(kNN, self).__repr__(
            ["k=%d" % self.__k, "dfx=%s" % self.__dfx,
             "voting=%s" % repr(self.__voting)]
            + prefixes)


    def __str__(self):
        return "%s\n data: %s" % \
            (Classifier.__str__(self), indentDoc(self.__data))


    def _train(self, data):
        """Train the classifier.

        For kNN it is degenerate -- just stores the data.
        """
        self.__data = data
        if __debug__:
            if str(data.samples.dtype).startswith('uint') \
                or str(data.samples.dtype).startswith('int'):
                warning("kNN: input data is in integers. " + \
                        "Overflow on arithmetic operations might result in"+\
                        " errors. Please convert dataset's samples into" +\
                        " floating datatype if any error is reported.")
        self.__weights = None

        # create dictionary with an item for each condition
        uniquelabels = data.uniquelabels
        self.__votes_init = dict(zip(uniquelabels,
                                     [0] * len(uniquelabels)))


    def _predict(self, data):
        """Predict the class labels for the provided data.

        Returns a list of class labels (one for each data sample).
        """
        # make sure we're talking about arrays
        data = N.asarray(data)

        # checks only in debug mode
        if __debug__:
            if not data.ndim == 2:
                raise ValueError, "Data array must be two-dimensional."

            if not data.shape[1] == self.__data.nfeatures:
                raise ValueError, "Length of data samples (features) does " \
                                  "not match the classifier."

        # compute the distance matrix between training and test data with
        # distances stored row-wise, ie. distances between test sample [0]
        # and all training samples will end up in row 0
        dists = self.__dfx(self.__data.samples, data).T

        # determine the k nearest neighbors per test sample
        knns = dists.argsort(axis=1)[:, :self.__k]

        # predicted class labels will go here
        predicted = []

        if self.__voting == 'majority':
            vfx = self.getMajorityVote
        elif self.__voting == 'weighted':
            vfx = self.getWeightedVote
        else:
            raise ValueError, "kNN told to perform unknown voting '%s'." \
                  % self.__voting

        # perform voting
        results = [vfx(knn) for knn in knns]

        # extract predictions
        predicted = [r[0] for r in results]

        # store the predictions in the state. Relies on State._setitem to do
        # nothing if the relevant state member is not enabled
        self.predictions = predicted
        self.values = [r[1] for r in results]

        return predicted


    def getMajorityVote(self, knn_ids):
        """Simple voting by choosing the majority of class neighbors.
        """
        # local bindings
        _data = self.__data
        labels = _data.labels

        # number of occerences for each unique class in kNNs
        votes = self.__votes_init.copy()
        for nn in knn_ids:
            votes[labels[nn]] += 1

        # find the class with most votes
        # return votes as well to store them in the state
        if _dict_has_key:
            # approx 5% faster implementation than below
            maxvotes = max(votes.iteritems(), key=lambda x:x[1])[0]
        else:
            # no key keyword for max in elderly versions
            maxvotes = max([(v, k) for k, v in votes.iteritems()])[1]

        return maxvotes, \
                [votes[ul] for ul in _data.uniquelabels] # transform into lists


    def getWeightedVote(self, knn_ids):
        """Vote with classes weighted by the number of samples per class.
        """
        # local bindings
        _data = self.__data
        uniquelabels = _data.uniquelabels

        # Lazy evaluation
        if self.__weights is None:
            #
            # It seemed to Yarik that this has to be evaluated just once per
            # training dataset.
            #
            self.__labels = labels = self.__data.labels
            Nlabels = len(labels)
            Nuniquelabels = len(uniquelabels)

            # TODO: To get proper speed up for the next line only,
            #       histogram should be computed
            #       via sorting + counting "same" elements while reducing.
            #       Guaranteed complexity is NlogN whenever now it is N^2
            # compute the relative proportion of samples belonging to each
            # class (do it in one loop to improve speed and reduce readability
            self.__weights = \
                [ 1.0 - ((labels == label).sum() / Nlabels) \
                    for label in uniquelabels ]
            self.__weights = dict(zip(uniquelabels, self.__weights))

        labels = self.__labels
        # number of occerences for each unique class in kNNs
        votes = self.__votes_init.copy()
        for nn in knn_ids:
            votes[labels[nn]] += 1

        # weight votes
        votes = [ self.__weights[ul] * votes[ul] for ul in uniquelabels]

        # find the class with most votes
        # return votes as well to store them in the state
        return uniquelabels[N.asarray(votes).argmax()], \
               votes


    def untrain(self):
        """Reset trained state"""
        self.__data = None
        super(kNN, self).untrain()