def residuals(data, layout):
"""Return the residuals for the given data and layout.
>>> residuals(np.array([1, 2, 3, 6], float), [[0, 1], [2, 3]])
array([-0.5, 0.5, -1.5, 1.5])
:param data:
An ndarray. Any number of dimensions is allowed.
:param layout:
A :term:`layout` describing the data.
:return:
The residuals obtained by subtracting the means of the groups
defined by the layout from the values in data.
"""
means = group_means(data, layout)
diffs = []
groups = apply_layout(data, layout)
for i, group in enumerate(groups):
these_means = means[..., i].reshape(np.shape(group)[:-1] + (1, ))
diffs.append(group - these_means)
return np.concatenate(diffs, axis=-1)
def residuals(data, layout):
"""Return the residuals for the given data and layout.
>>> residuals(np.array([1, 2, 3, 6], float), [[0, 1], [2, 3]])
array([-0.5, 0.5, -1.5, 1.5])
:param data:
An ndarray. Any number of dimensions is allowed.
:param layout:
A :term:`layout` describing the data.
:return:
The residuals obtained by subtracting the means of the groups
defined by the layout from the values in data.
"""
means = group_means(data, layout)
diffs = []
groups = apply_layout(data, layout)
for i, group in enumerate(groups):
these_means = means[..., i].reshape(np.shape(group)[:-1] + (1,))
diffs.append(group - these_means)
return np.concatenate(diffs, axis=-1)
def __call__(self, data):
groups = apply_layout(data, self.condition_layout)
res = []
for i, grp in enumerate(groups):
syms = []
for index in grp:
sym = self.index_to_symbol[index]
syms.append(sym)
res.append(''.join(sorted(syms)))
return ' '.join(res)
def __call__(self, data):
groups = apply_layout(data, self.condition_layout)
res = []
for i, grp in enumerate(groups):
syms = []
for index in grp:
sym = self.index_to_symbol[index]
syms.append(sym)
res.append(''.join(sorted(syms)))
return ' '.join(res)
def group_means(data, layout):
"""Get the means for each group defined by layout.
Groups data according to the given layout and returns a new
ndarray with the same number of dimensions as data, but with the
last dimension replaced by the means according to the specified
grouping.
One dimensional input:
>>> group_means(np.array([-1, -3, 4, 6]), [[0, 1], [2, 3]])
array([-2., 5.])
Two dimensional input:
>>> data = np.array([[-1, -3, 4, 6], [10, 12, 30, 40]])
>>> layout = [[0, 1], [2, 3]]
>>> group_means(data, layout) # doctest: +NORMALIZE_WHITESPACE
array([[ -2., 5.],
[ 11., 35.]])
:param data: An ndarray. Any number of dimensions is allowed.
:param layout: A :term:`layout` describing the data.
:return: An ndarray giving the means for each group obtained by
applying the given layout to the given data.
"""
# We'll take the mean of the last axis of each group, so change
# the shape of the array to collapse the last axis down to one
# item per group.
shape = np.shape(data)[:-1] + (len(layout), )
res = np.zeros(shape)
for i, group in enumerate(apply_layout(data, layout)):
res[..., i] = np.mean(group, axis=-1)
return res
def group_means(data, layout):
"""Get the means for each group defined by layout.
Groups data according to the given layout and returns a new
ndarray with the same number of dimensions as data, but with the
last dimension replaced by the means according to the specified
grouping.
One dimensional input:
>>> group_means(np.array([-1, -3, 4, 6]), [[0, 1], [2, 3]])
array([-2., 5.])
Two dimensional input:
>>> data = np.array([[-1, -3, 4, 6], [10, 12, 30, 40]])
>>> layout = [[0, 1], [2, 3]]
>>> group_means(data, layout) # doctest: +NORMALIZE_WHITESPACE
array([[ -2., 5.],
[ 11., 35.]])
:param data: An ndarray. Any number of dimensions is allowed.
:param layout: A :term:`layout` describing the data.
:return: An ndarray giving the means for each group obtained by
applying the given layout to the given data.
"""
# We'll take the mean of the last axis of each group, so change
# the shape of the array to collapse the last axis down to one
# item per group.
shape = np.shape(data)[:-1] + (len(layout),)
res = np.zeros(shape)
for i, group in enumerate(apply_layout(data, layout)):
res[..., i] = np.mean(group, axis=-1)
return res
def sep(self, data):
data = np.asarray(data)
groups = apply_layout(data, self.condition_layout)
return np.abs(
np.mean(groups[0], axis=-1) - np.mean(groups[1], axis=-1))
def sep(self, data):
data = np.asarray(data)
groups = apply_layout(data, self.condition_layout)
return np.abs(np.mean(groups[0], axis=-1) - np.mean(groups[1], axis=-1))