def sort(self, data, index=None):
"""Applies a (potentially hierarchical) sort operation on a numpy array
or pandas series/dataframe based on the grouping index or a
user-supplied index. Returns an object of the same type as the
original data as well as the matching (sorted) Pandas index.
"""
if index is None:
index = self.index
if data_util._is_using_ndarray_type(data, None):
if data.ndim == 1:
out = pd.Series(data, index=index, copy=True)
out = out.sort_index()
else:
out = pd.DataFrame(data, index=index)
out = out.sort_index(inplace=False) # copies
return np.array(out), out.index
elif data_util._is_using_pandas(data, None):
out = data
out = out.reindex(index) # copies?
out = out.sort_index()
return out, out.index
else:
msg = 'data must be a Numpy array or a Pandas Series/DataFrame'
raise ValueError(msg)
def handle_data(endog, exog, missing='none', hasconst=None, **kwargs):
"""
Given inputs
"""
# deal with lists and tuples up-front
if isinstance(endog, (list, tuple)):
endog = np.asarray(endog)
if isinstance(exog, (list, tuple)):
exog = np.asarray(exog)
if data_util._is_using_ndarray_type(endog, exog):
klass = ModelData
elif data_util._is_using_pandas(endog, exog):
klass = PandasData
elif data_util._is_using_patsy(endog, exog):
klass = PatsyData
# keep this check last
elif data_util._is_using_ndarray(endog, exog):
klass = ModelData
else:
raise ValueError('unrecognized data structures: %s / %s' %
(type(endog), type(exog)))
return klass(endog,
exog=exog,
missing=missing,
hasconst=hasconst,
**kwargs)
def sort(self, data, index=None):
'''Applies a (potentially hierarchical) sort operation on a numpy array
or pandas series/dataframe based on the grouping index or a
user-supplied index. Returns an object of the same type as the
original data as well as the matching (sorted) Pandas index.
'''
if index is None:
index = self.index
if data_util._is_using_ndarray_type(data, None):
if data.ndim == 1:
out = pd.Series(data, index=index, copy=True)
out = out.sort_index()
else:
out = pd.DataFrame(data, index=index)
out = out.sort(inplace=False) # copies
return np.array(out), out.index
elif data_util._is_using_pandas(data, None):
out = data
out = out.reindex(index) # copies?
out = out.sort_index()
return out, out.index
else:
msg = 'data must be a Numpy array or a Pandas Series/DataFrame'
raise ValueError(msg)
def handle_data(endog, exog):
"""
Given inputs
"""
# deal with lists and tuples up-front
if isinstance(endog, (list, tuple)):
endog = np.asarray(endog)
if isinstance(exog, (list, tuple)):
exog = np.asarray(exog)
if data_util._is_using_ndarray_type(endog, exog):
klass = ModelData
elif data_util._is_using_pandas(endog, exog):
klass = PandasData
elif data_util._is_using_larry(endog, exog):
klass = LarryData
elif data_util._is_using_timeseries(endog, exog):
klass = TimeSeriesData
elif data_util._is_using_patsy(endog, exog):
klass = PatsyData
# keep this check last
elif data_util._is_using_ndarray(endog, exog):
klass = ModelData
else:
raise ValueError("unrecognized data structures: %s / %s" % (type(endog), type(exog)))
return klass(endog, exog=exog)
def group_sums_dummy(x, group_dummy):
"""sum by groups given group dummy variable
group_dummy can be either ndarray or sparse matrix
"""
if data_util._is_using_ndarray_type(group_dummy, None):
return np.dot(x.T, group_dummy)
else: # check for sparse
return x.T * group_dummy
def group_sums_dummy(x, group_dummy):
'''sum by groups given group dummy variable
group_dummy can be either ndarray or sparse matrix
'''
if data_util._is_using_ndarray_type(group_dummy, None):
return np.dot(x.T, group_dummy)
else: # check for sparse
return x.T * group_dummy
def handle_data_class_factory(endog, exog):
"""
Given inputs
"""
if data_util._is_using_ndarray_type(endog, exog):
klass = ModelData
elif data_util._is_using_pandas(endog, exog):
klass = PandasData
elif data_util._is_using_patsy(endog, exog):
klass = PatsyData
# keep this check last
elif data_util._is_using_ndarray(endog, exog):
klass = ModelData
else:
raise ValueError("unrecognized data structures: %s / %s" % (type(endog), type(exog)))
return klass
def handle_data_class_factory(endog, exog):
"""
Given inputs
"""
if data_util._is_using_ndarray_type(endog, exog):
klass = ModelData
elif data_util._is_using_pandas(endog, exog):
klass = PandasData
elif data_util._is_using_patsy(endog, exog):
klass = PatsyData
# keep this check last
elif data_util._is_using_ndarray(endog, exog):
klass = ModelData
else:
raise ValueError('unrecognized data structures: %s / %s' %
(type(endog), type(exog)))
return klass
def handle_data(endog, exog, missing="none", hasconst=None, **kwargs):
"""
Given inputs
"""
# deal with lists and tuples up-front
if isinstance(endog, (list, tuple)):
endog = np.asarray(endog)
if isinstance(exog, (list, tuple)):
exog = np.asarray(exog)
if data_util._is_using_ndarray_type(endog, exog):
klass = ModelData
elif data_util._is_using_pandas(endog, exog):
klass = PandasData
elif data_util._is_using_patsy(endog, exog):
klass = PatsyData
# keep this check last
elif data_util._is_using_ndarray(endog, exog):
klass = ModelData
else:
raise ValueError("unrecognized data structures: %s / %s" % (type(endog), type(exog)))
return klass(endog, exog=exog, missing=missing, hasconst=hasconst, **kwargs)
