def train(self, data, **args):
# store the current cpu time:
self._clock = time.clock()
if not data.labels.numericLabels:
# check if there is a class that is not represented in the training data:
if min(data.labels.classSize) == 0:
raise ValueError, 'there is a class with no data'
# store just as much about the labels as is needed:
self.labels = misc.Container()
self.labels.addAttributes(data.labels,
['numClasses', 'classLabels'])
# if dealing with a VectorDataSet test data needs to have the same features
if data.__class__.__name__ == 'VectorDataSet':
self.featureID = data.featureID[:]
data.train(**args)
# if there is some testing done on the data, it requires the training data:
if data.testingFunc is not None:
self.trainingData = data
def convert (object, attributes) :
obj = misc.Container()
obj.addAttributes(object, attributes)
return obj
def __init__(self, arg=None, **args):
PyMLobject.__init__(self, arg, **args)
if type(arg) == type(''):
self.load(arg)
self.log = misc.Container()