def test_ctor_dict(self):
itema = self.panel["ItemA"]
itemb = self.panel["ItemB"]
d = {"A": itema, "B": itemb[5:]}
d2 = {"A": itema._series, "B": itemb[5:]._series}
d3 = {"A": None, "B": DataFrame(itemb[5:]._series), "C": DataFrame(itema._series)}
wp = Panel.from_dict(d)
wp2 = Panel.from_dict(d2) # nested Dict
wp3 = Panel.from_dict(d3)
self.assert_(wp.major_axis.equals(self.panel.major_axis))
assert_panel_equal(wp, wp2)
# intersect
wp = Panel.from_dict(d, intersect=True)
self.assert_(wp.major_axis.equals(itemb.index[5:]))
# use constructor
assert_panel_equal(Panel(d), Panel.from_dict(d))
assert_panel_equal(Panel(d2), Panel.from_dict(d2))
assert_panel_equal(Panel(d3), Panel.from_dict(d3))
# a pathological case
d4 = {"A": None, "B": None}
wp4 = Panel.from_dict(d4)
assert_panel_equal(Panel(d4), Panel(items=["A", "B"]))
# cast
dcasted = dict((k, v.reindex(wp.major_axis).fillna(0)) for k, v in d.iteritems())
result = Panel(dcasted, dtype=int)
expected = Panel(dict((k, v.astype(int)) for k, v in dcasted.iteritems()))
assert_panel_equal(result, expected)
def test_shift(self):
# mixed dtypes #6959
data = [('item ' + ch, makeMixedDataFrame()) for ch in list('abcde')]
data = dict(data)
mixed_panel = Panel.from_dict(data, orient='minor')
shifted = mixed_panel.shift(1)
assert_series_equal(mixed_panel.dtypes, shifted.dtypes)
def test_ctor_dict(self):
itema = self.panel['ItemA']
itemb = self.panel['ItemB']
d = {'A' : itema, 'B' : itemb[5:]}
d2 = {'A' : itema._series, 'B' : itemb[5:]._series}
d3 = {'A' : DataFrame(itema._series),
'B' : DataFrame(itemb[5:]._series)}
wp = Panel.from_dict(d)
wp2 = Panel.from_dict(d2) # nested Dict
wp3 = Panel.from_dict(d3)
self.assert_(wp.major_axis.equals(self.panel.major_axis))
assert_panel_equal(wp, wp2)
# intersect
wp = Panel.from_dict(d, intersect=True)
self.assert_(wp.major_axis.equals(itemb.index[5:]))
# use constructor
assert_panel_equal(Panel(d), Panel.from_dict(d))
assert_panel_equal(Panel(d2), Panel.from_dict(d2))
assert_panel_equal(Panel(d3), Panel.from_dict(d3))
# cast
result = Panel(d, dtype=int)
expected = Panel(dict((k, v.astype(int)) for k, v in d.iteritems()))
def test_ctor_dict(self):
itema = self.panel['ItemA']
itemb = self.panel['ItemB']
d = {'A': itema, 'B': itemb[5:]}
d2 = {'A': itema._series, 'B': itemb[5:]._series}
d3 = {'A': DataFrame(itema._series), 'B': DataFrame(itemb[5:]._series)}
wp = Panel.from_dict(d)
wp2 = Panel.from_dict(d2) # nested Dict
wp3 = Panel.from_dict(d3)
self.assert_(wp.major_axis.equals(self.panel.major_axis))
assert_panel_equal(wp, wp2)
# intersect
wp = Panel.from_dict(d, intersect=True)
self.assert_(wp.major_axis.equals(itemb.index[5:]))
# use constructor
assert_panel_equal(Panel(d), Panel.from_dict(d))
assert_panel_equal(Panel(d2), Panel.from_dict(d2))
assert_panel_equal(Panel(d3), Panel.from_dict(d3))
# cast
result = Panel(d, dtype=int)
expected = Panel(dict((k, v.astype(int)) for k, v in d.iteritems()))
def test_ctor_orderedDict(self):
keys = list(set(np.random.randint(
0, 5000, 100)))[:50] # unique random int keys
d = OrderedDict([(k, mkdf(10, 5)) for k in keys])
p = Panel(d)
assert list(p.items) == keys
p = Panel.from_dict(d)
assert list(p.items) == keys
def var_beta(self):
"""Returns the covariance of beta."""
result = {}
result_index = self._result_index
for i in range(len(self._var_beta_raw)):
dm = DataFrame(self._var_beta_raw[i], columns=self.beta.columns, index=self.beta.columns)
result[result_index[i]] = dm
return Panel.from_dict(result, intersect=False)
def test_from_dict_mixed_orient(self):
df = tm.makeDataFrame()
df["foo"] = "bar"
data = {"k1": df, "k2": df}
panel = Panel.from_dict(data, orient="minor")
self.assert_(panel["foo"].values.dtype == np.object_)
self.assert_(panel["A"].values.dtype == np.float64)
def test_from_dict_mixed_orient(self):
df = tm.makeDataFrame()
df['foo'] = 'bar'
data = {'k1': df, 'k2': df}
panel = Panel.from_dict(data, orient='minor')
self.assert_(panel['foo'].values.dtype == np.object_)
self.assert_(panel['A'].values.dtype == np.float64)
def var_beta(self):
"""Returns the covariance of beta."""
result = {}
result_index = self._result_index
for i in xrange(len(self._var_beta_raw)):
dm = DataFrame(self._var_beta_raw[i], columns=self.beta.columns,
index=self.beta.columns)
result[result_index[i]] = dm
return Panel.from_dict(result, intersect=False)
def test_from_dict_mixed_orient(self):
df = tm.makeDataFrame()
df['foo'] = 'bar'
data = {'k1' : df,
'k2' : df}
panel = Panel.from_dict(data, orient='minor')
self.assert_(panel['foo'].values.dtype == np.object_)
self.assert_(panel['A'].values.dtype == np.float64)
def _filter_data(self):
"""
"""
data = self._x_orig
cat_mapping = {}
if isinstance(data, LongPanel):
data = data.to_wide()
else:
if isinstance(data, Panel):
data = data.copy()
if not isinstance(data, SparsePanel):
data, cat_mapping = self._convert_x(data)
if not isinstance(data, Panel):
data = Panel.from_dict(data, intersect=True)
x_names = data.items
if self._weights is not None:
data['__weights__'] = self._weights
# Filter x's without y (so we can make a prediction)
filtered = data.to_long()
# Filter all data together using to_long
# convert to DataFrame
y = self._y_orig
if isinstance(y, Series):
y = y.unstack()
data['__y__'] = y
data_long = data.to_long()
x_filt = filtered.filter(x_names)
if self._weights:
weights_filt = filtered['__weights__']
else:
weights_filt = None
x = data_long.filter(x_names)
y = data_long['__y__']
if self._weights:
weights = data_long['__weights__']
else:
weights = None
return x, x_filt, y, weights, weights_filt, cat_mapping
def test_constructor_dtypes(self):
# GH #797
def _check_dtype(panel, dtype):
for i in panel.items:
assert panel[i].values.dtype.name == dtype
# only nan holding types allowed here
for dtype in ['float64', 'float32', 'object']:
panel = Panel(items=lrange(2),
major_axis=lrange(10),
minor_axis=lrange(5),
dtype=dtype)
_check_dtype(panel, dtype)
for dtype in ['float64', 'float32', 'int64', 'int32', 'object']:
panel = Panel(np.array(np.random.randn(2, 10, 5), dtype=dtype),
items=lrange(2),
major_axis=lrange(10),
minor_axis=lrange(5),
dtype=dtype)
_check_dtype(panel, dtype)
for dtype in ['float64', 'float32', 'int64', 'int32', 'object']:
panel = Panel(np.array(np.random.randn(2, 10, 5), dtype='O'),
items=lrange(2),
major_axis=lrange(10),
minor_axis=lrange(5),
dtype=dtype)
_check_dtype(panel, dtype)
for dtype in ['float64', 'float32', 'int64', 'int32', 'object']:
panel = Panel(np.random.randn(2, 10, 5),
items=lrange(2),
major_axis=lrange(10),
minor_axis=lrange(5),
dtype=dtype)
_check_dtype(panel, dtype)
for dtype in ['float64', 'float32', 'int64', 'int32', 'object']:
df1 = DataFrame(np.random.randn(2, 5),
index=lrange(2),
columns=lrange(5))
df2 = DataFrame(np.random.randn(2, 5),
index=lrange(2),
columns=lrange(5))
panel = Panel.from_dict({'a': df1, 'b': df2}, dtype=dtype)
_check_dtype(panel, dtype)
def _filter_data(self):
"""
"""
data = self._x_orig
cat_mapping = {}
if isinstance(data, DataFrame):
data = data.to_panel()
else:
if isinstance(data, Panel):
data = data.copy()
if not isinstance(data, SparsePanel):
data, cat_mapping = self._convert_x(data)
if not isinstance(data, Panel):
data = Panel.from_dict(data, intersect=True)
x_names = data.items
if self._weights is not None:
data['__weights__'] = self._weights
# Filter x's without y (so we can make a prediction)
filtered = data.to_frame()
# Filter all data together using to_frame
# convert to DataFrame
y = self._y_orig
if isinstance(y, Series):
y = y.unstack()
data['__y__'] = y
data_long = data.to_frame()
x_filt = filtered.filter(x_names)
x = data_long.filter(x_names)
y = data_long['__y__']
if self._weights is not None and not self._weights.empty:
weights = data_long['__weights__']
else:
weights = None
return x, x_filt, y, weights, cat_mapping