def construct(self, data = None, objectClass = None):
"""
construct() creates data structures -- attributes or examples -- as indicated by the obj variable. Returns an AttributeSet or ExampleSet data structure.
data: resource required to construct object.
objectClass: indicator for the type of object class to use. Signals construction of ExampleSet.
"""
if data is None:
raise ValueError("No data specified.")
elif objectClass is None:
a = AttributeSet()
for line in data:
a.add(Attribute(line.split('\t')))
return a
elif objectClass is not None:
e = ExampleSet()
for line in data:
e.add(Example(line, objectClass))
return e
else:
raise ValueError("Object type needs to be indicated as either 'Attribute' (0) or 'Example' (1).")
def build(self, data = None, attributeSet = None):
if attributeSet is None:
# build an attribute set.
aa = AttributeSet()
for line in data:
if line[0] == '@':
aa.add(Attribute(line[1:].split('\t')))
return aa
else:
# build an example set.
ee = ExampleSet()
for line in data:
if type(line) == type(str()):
if line[0] == '#':
ee.add(Example(line[1:], attributeSet))
elif type(line) == type(ExampleSet()):
ee.add(line)
return ee
def __init__(self, filePath=None): self.name = None self.attributes = None self.examples = ExampleSet() self.iteration_index = 0 if filePath is not None: self.initialize(filePath)
def getTrainValidateTestSet(self, p = .6, v = .5):
examples = self.getExamples()
n = int(len(examples) * p)
m = int(len(examples) * ((1. - p)*v))
s = sample(examples, n)
train = ExampleSet()
valid = ExampleSet()
tests = ExampleSet()
for example in examples:
if example in s: train.add(example)
elif example not in train and m != 0: valid.add(example); m-=1
elif example not in valid: tests.add(example)
print "train: {0} valid: {1} tests: {2} all: {3}".format(len(train), len(valid), len(tests), len(self.getExamples()))
return train, valid, tests
def build(self, data=None, attributeSet=None):
""" return an AttributeSet or ExampleSet object
@param data: input data; raw (textual) attributes or examples
@param attributeSet: AttributeSet object required to create ExampleSet objects
@return AttributeSet or ExampleSet objects
"""
# Build an AttributeSet object from raw (text) attributes.
if attributeSet is None:
attributeSet = AttributeSet()
for line in data:
# If the line is prefixed with '@', create an Attribute object and add it to the AttributeSet
if line[0] == '@':
attributeSet.add(Attribute(line[1:].split('\t')))
return attributeSet
# Build an ExampleSet object from raw (text) examples and an AttributeSet.
else:
exampleSet = ExampleSet()
# Loop through the data split by newline
for line in data:
# If the line is a string, check it is an example (prefixed by '#')
if type(line) == type(str()):
# If the line is an example, create an Example object and add it to the ExampleSet
if line[0] == '#':
exampleSet.add(Example(line[1:], attributeSet))
# Commented out for the time being 7/13/2016
#else:
# exampleSet.add(line)
return exampleSet
def build(self, data = None, attributeSet = None):
""" return an AttributeSet or ExampleSet object
@param data: input data; raw (textual) attributes or examples
@param attributeSet: AttributeSet object required to create ExampleSet objects
@return AttributeSet or ExampleSet objects
"""
# Build an AttributeSet object from raw (text) attributes.
if attributeSet is None:
attributeSet = AttributeSet()
for line in data:
# If the line is prefixed with '@', create an Attribute object and add it to the AttributeSet
if line[0] == '@':
attributeSet.add(Attribute(line[1:].split('\t')))
return attributeSet
# Build an ExampleSet object from raw (text) examples and an AttributeSet.
else:
exampleSet = ExampleSet()
# Loop through the data split by newline
for line in data:
# If the line is a string, check it is an example (prefixed by '#')
if type(line) == type(str()):
# If the line is an example, create an Example object and add it to the ExampleSet
if line[0] == '#':
exampleSet.add(Example(line[1:], attributeSet))
# Commented out for the time being 7/13/2016
#else:
# exampleSet.add(line)
return exampleSet
def getTrainTestSet(self, p = .6): examples = self.getExamples() n = int(len(examples) * p) s = sample(examples, n) train = ExampleSet() tests = ExampleSet() for example in examples: if example in s: train.add(example) elif example not in train: tests.add(example) return train, tests
def getExamplesByClass(self, i = None): """ return examples with label i """ return ExampleSet(self.examples.getExamples(i))
