def _curve_wraps(self, func, *args, **kwargs):
decision = self._df.decision
if util._is_1d_varray(decision):
c1, c2, threshold = func(self._target.values, decision.values,
*args, **kwargs)
return c1, c2, threshold
results = {}
if self._df.has_multi_targets():
for i, ((name, col), t) in enumerate(
zip(decision.iteritems(), self._target.values.T)):
# results can have different length
c1, c2, threshold = func(t,
col.values,
pos_label=i,
*args,
**kwargs)
results[name] = c1, c2, threshold
else:
for i, (name, col) in enumerate(decision.iteritems()):
# results can have different length
c1, c2, threshold = func(self._target.values,
col.values,
pos_label=i,
*args,
**kwargs)
results[name] = c1, c2, threshold
return results
def _wrap_predicted(self, predicted, estimator):
"""
Wrapper for predict methods
"""
if util._is_1d_varray(predicted):
predicted = self._constructor_sliced(predicted, index=self.index)
else:
predicted = self._constructor(predicted, index=self.index)
self._predicted = predicted
return self._predicted
def _wrap_predicted(self, predicted, estimator):
"""
Wrapper for predict methods
"""
if util._is_1d_varray(predicted):
predicted = self._constructor_sliced(predicted, index=self.index)
else:
predicted = self._constructor(predicted, index=self.index)
self._predicted = predicted
return self._predicted
def _wrap_transform(self, transformed, columns=None):
"""
Wrapper for transform methods
"""
if len(transformed.shape) == 2:
if (util._is_1d_harray(transformed)
or util._is_1d_varray(transformed)):
transformed = transformed.flatten()
else:
from pandas_ml.core.frame import ModelFrame
return ModelFrame(transformed, index=self.index)
return self._constructor(transformed, index=self.index, name=self.name)
def _wrap_transform(self, transformed, columns=None):
"""
Wrapper for transform methods
"""
if len(transformed.shape) == 2:
if (util._is_1d_harray(transformed) or
util._is_1d_varray(transformed)):
transformed = transformed.flatten()
else:
from pandas_ml.core.frame import ModelFrame
return ModelFrame(transformed, index=self.index)
return self._constructor(transformed, index=self.index,
name=self.name)
def _wrap_probability(self, probability, estimator):
"""
Wrapper for probability methods
"""
try:
if util._is_1d_varray(probability):
# 2 class
probability = self._constructor(probability, index=self.index)
else:
probability = self._constructor(probability, index=self.index,
columns=estimator.classes_)
except ValueError:
msg = "Unable to instantiate ModelFrame for '{0}'"
warnings.warn(msg.format(estimator.__class__.__name__))
return probability
def _wrap_probability(self, probability, estimator):
"""
Wrapper for probability methods
"""
try:
if util._is_1d_varray(probability):
# 2 class
probability = self._constructor(probability, index=self.index)
else:
probability = self._constructor(probability, index=self.index,
columns=estimator.classes_)
except ValueError:
msg = "Unable to instantiate ModelFrame for '{0}'"
warnings.warn(msg.format(estimator.__class__.__name__))
return probability
def _curve_wraps(self, func, *args, **kwargs):
decision = self._df.decision
if util._is_1d_varray(decision):
c1, c2, threshold = func(self._target.values, decision.values,
*args, **kwargs)
return c1, c2, threshold
results = {}
if self._df.has_multi_targets():
for i, ((name, col), t) in enumerate(zip(decision.iteritems(), self._target.values.T)):
# results can have different length
c1, c2, threshold = func(t, col.values, pos_label=i, *args, **kwargs)
results[name] = c1, c2, threshold
else:
for i, (name, col) in enumerate(decision.iteritems()):
# results can have different length
c1, c2, threshold = func(self._target.values, col.values,
pos_label=i, *args, **kwargs)
results[name] = c1, c2, threshold
return results
