def summary(self, xname=None, alpha=0.05, title=None):
if self.effect is not None:
# TODO: should also add some extra information, e.g. robust cov ?
# TODO: can we infer names for constraints, xname in __init__ ?
if title is None:
title = 'Test for Constraints'
elif title == '':
# don't add any title,
# I think SimpleTable skips on None - check
title = None
# we have everything for a params table
use_t = (self.distribution == 't')
yname='constraints' # Not used in params_frame
if xname is None:
xname = ['c%d'%ii for ii in range(len(self.effect))]
from statsmodels.iolib.summary import summary_params
summ = summary_params((self, self.effect, self.sd, self.statistic,
self.pvalue, self.conf_int(alpha)),
yname=yname, xname=xname, use_t=use_t,
title=title)
return summ
elif hasattr(self, 'fvalue'):
# TODO: create something nicer for these casee
return '<F test: F=%s, p=%s, df_denom=%d, df_num=%d>' % \
(repr(self.fvalue), self.pvalue, self.df_denom, self.df_num)
else:
# generic
return '<Wald test: statistic=%s, p-value=%s>' % \
(self.statistic, self.pvalue)
def summary(self, xname=None, alpha=0.05, title=None):
"""Summarize the Results of the hypothesis test
Parameters
-----------
xname : list of strings, optional
Default is `c_##` for ## in p the number of regressors
alpha : float
significance level for the confidence intervals. Default is
alpha = 0.05 which implies a confidence level of 95%.
title : string, optional
Title for the params table. If not None, then this replaces the
default title
Returns
-------
smry : string or Summary instance
This contains a parameter results table in the case of t or z test
in the same form as the parameter results table in the model
results summary.
For F or Wald test, the return is a string.
"""
if self.effect is not None:
# TODO: should also add some extra information, e.g. robust cov ?
# TODO: can we infer names for constraints, xname in __init__ ?
if title is None:
title = 'Test for Constraints'
elif title == '':
# don't add any title,
# I think SimpleTable skips on None - check
title = None
# we have everything for a params table
use_t = (self.distribution == 't')
yname = 'constraints' # Not used in params_frame
if xname is None:
xname = ['c%d' % ii for ii in range(len(self.effect))]
from statsmodels.iolib.summary import summary_params
pvalues = np.atleast_1d(self.pvalue)
summ = summary_params(
(self, self.effect, self.sd, self.statistic, pvalues,
self.conf_int(alpha)),
yname=yname,
xname=xname,
use_t=use_t,
title=title,
alpha=alpha)
return summ
elif hasattr(self, 'fvalue'):
# TODO: create something nicer for these casee
return '<F test: F=%s, p=%s, df_denom=%d, df_num=%d>' % \
(repr(self.fvalue), self.pvalue, self.df_denom, self.df_num)
elif self.distribution == 'chi2':
return '<Wald test (%s): statistic=%s, p-value=%s, df_denom=%d>' % \
(self.distribution, self.statistic, self.pvalue, self.df_denom)
else:
# generic
return '<Wald test: statistic=%s, p-value=%s>' % \
(self.statistic, self.pvalue)
def summary(self, use_t=True, alpha=0.05, usevar='pooled', value=0):
'''summarize the results of the hypothesis test
Parameters
----------
use_t : bool, optional
if use_t is True, then t test results are returned
if use_t is False, then z test results are returned
alpha : float
significance level for the confidence interval, coverage is
``1-alpha``
usevar : string, 'pooled' or 'unequal'
If ``pooled``, then the standard deviation of the samples is
assumed to be the same. If ``unequal``, then the variance of
Welsh ttest will be used, and the degrees of freedom are those
of Satterthwaite if ``use_t`` is True.
value : float
difference between the means under the Null hypothesis.
Returns
-------
smry : SimpleTable
'''
d1 = self.d1
d2 = self.d2
confint_percents = 100 - alpha * 100
if use_t:
tstat, pvalue, _ = self.ttest_ind(usevar=usevar, value=value)
lower, upper = self.tconfint_diff(alpha=alpha, usevar=usevar)
else:
tstat, pvalue = self.ztest_ind(usevar=usevar, value=value)
lower, upper = self.zconfint_diff(alpha=alpha, usevar=usevar)
if usevar == 'pooled':
std_err = self.std_meandiff_pooledvar
else:
std_err = self.std_meandiff_separatevar
std_err = np.atleast_1d(std_err)
tstat = np.atleast_1d(tstat)
pvalue = np.atleast_1d(pvalue)
lower = np.atleast_1d(lower)
upper = np.atleast_1d(upper)
conf_int = np.column_stack((lower, upper))
params = d1.mean - d2.mean - value
title = 'Test for equality of means'
yname = 'y' # not used in params_frame
xname = ['subset #%d'%(ii + 1) for ii in range(tstat.shape[0])]
from statsmodels.iolib.summary import summary_params
return summary_params((None, params, std_err, tstat, pvalue, conf_int),
alpha=alpha, use_t=use_t, yname=yname, xname=xname,
title=title)
def summary(self, use_t=True, alpha=0.05, usevar='pooled', value=0):
'''summarize the results of the hypothesis test
Parameters
----------
use_t : bool, optional
if use_t is True, then t test results are returned
if use_t is False, then z test results are returned
alpha : float
significance level for the confidence interval, coverage is
``1-alpha``
usevar : string, 'pooled' or 'unequal'
If ``pooled``, then the standard deviation of the samples is
assumed to be the same. If ``unequal``, then the variance of
Welsh ttest will be used, and the degrees of freedom are those
of Satterthwaite if ``use_t`` is True.
value : float
difference between the means under the Null hypothesis.
Returns
-------
smry : SimpleTable
'''
d1 = self.d1
d2 = self.d2
confint_percents = 100 - alpha * 100
if use_t:
tstat, pvalue, _ = self.ttest_ind(usevar=usevar, value=value)
lower, upper = self.tconfint_diff(alpha=alpha, usevar=usevar)
else:
tstat, pvalue = self.ztest_ind(usevar=usevar, value=value)
lower, upper = self.zconfint_diff(alpha=alpha, usevar=usevar)
if usevar == 'pooled':
std_err = self.std_meandiff_pooledvar
else:
std_err = self.std_meandiff_separatevar
std_err = np.atleast_1d(std_err)
tstat = np.atleast_1d(tstat)
pvalue = np.atleast_1d(pvalue)
lower = np.atleast_1d(lower)
upper = np.atleast_1d(upper)
conf_int = np.column_stack((lower, upper))
params = d1.mean - d2.mean - value
title = 'Test for equality of means'
yname = 'y' # not used in params_frame
xname = ['subset #%d'%(ii + 1) for ii in range(tstat.shape[0])]
from statsmodels.iolib.summary import summary_params
return summary_params((None, params, std_err, tstat, pvalue, conf_int),
alpha=alpha, use_t=use_t, yname=yname, xname=xname,
title=title)
def summary(self, xname=None, alpha=0.05, title=None):
"""Summarize the Results of the hypothesis test
Parameters
-----------
xname : list of strings, optional
Default is `c_##` for ## in p the number of regressors
alpha : float
significance level for the confidence intervals. Default is
alpha = 0.05 which implies a confidence level of 95%.
title : string, optional
Title for the params table. If not None, then this replaces the
default title
Returns
-------
smry : string or Summary instance
This contains a parameter results table in the case of t or z test
in the same form as the parameter results table in the model
results summary.
For F or Wald test, the return is a string.
"""
if self.effect is not None:
# TODO: should also add some extra information, e.g. robust cov ?
# TODO: can we infer names for constraints, xname in __init__ ?
if title is None:
title = 'Test for Constraints'
elif title == '':
# don't add any title,
# I think SimpleTable skips on None - check
title = None
# we have everything for a params table
use_t = (self.distribution == 't')
yname='constraints' # Not used in params_frame
if xname is None:
xname = ['c%d' % ii for ii in range(len(self.effect))]
from statsmodels.iolib.summary import summary_params
pvalues = np.atleast_1d(self.pvalue)
summ = summary_params((self, self.effect, self.sd, self.statistic,
pvalues, self.conf_int(alpha)),
yname=yname, xname=xname, use_t=use_t,
title=title, alpha=alpha)
return summ
elif hasattr(self, 'fvalue'):
# TODO: create something nicer for these casee
return ('<F test: F=%s, p=%s, df_denom=%.3g, df_num=%.3g>' %
(repr(self.fvalue), self.pvalue, self.df_denom,
self.df_num))
elif self.distribution == 'chi2':
return ('<Wald test (%s): statistic=%s, p-value=%s, df_denom=%.3g>' %
(self.distribution, self.statistic, self.pvalue,
self.df_denom))
else:
# generic
return ('<Wald test: statistic=%s, p-value=%s>' %
(self.statistic, self.pvalue))
def get_confidence_interval(self):
summary_data = summary_params(self.logit_result_linear).data
intercept_range = [
float(summary_data[1][5]),
float(summary_data[1][6])
]
slope_range = [float(summary_data[2][5]), float(summary_data[2][6])]
# print(intercept_range)
# print(slope_range)
return intercept_range, slope_range
def make_table(self, mask, title, strip_end=True):
res = (self, self.params[mask], self.bse[mask],
self.zvalues[mask], self.pvalues[mask],
self.conf_int(alpha)[mask])
param_names = [
'.'.join(name.split('.')[:-1]) if strip_end else name
for name in
np.array(self.data.param_names)[mask].tolist()
]
return summary_params(res, yname=None, xname=param_names,
alpha=alpha, use_t=False, title=title)
def make_table(self, mask, title, strip_end=True):
res = (self, self.params[mask], self.bse[mask],
self.zvalues[mask], self.pvalues[mask],
self.conf_int(alpha)[mask])
param_names = [
'.'.join(name.split('.')[:-1]) if strip_end else name
for name in
np.array(self.data.param_names)[mask].tolist()
]
return summary_params(res, yname=None, xname=param_names,
alpha=alpha, use_t=False, title=title)
def summary(self, alpha=0.05, xname=None):
"""summary table for probability that random draw x1 is larger than x2
Parameters
----------
alpha : float
Significance level for confidence intervals. Coverage is 1 - alpha
xname : None or list of str
If None, then each row has a name column with generic names.
If xname is a list of strings, then it will be included as part
of those names.
Returns
-------
SimpleTable instance with methods to convert to different output
formats.
"""
yname = "None"
effect = np.atleast_1d(self.prob1)
if self.pvalue is None:
statistic, pvalue = self.test_prob_superior()
else:
pvalue = self.pvalue
statistic = self.statistic
pvalues = np.atleast_1d(pvalue)
ci = np.atleast_2d(self.conf_int(alpha))
if ci.shape[0] > 1:
ci = ci.T
use_t = self.use_t
sd = np.atleast_1d(np.sqrt(self.var_prob))
statistic = np.atleast_1d(statistic)
if xname is None:
xname = ['c%d' % ii for ii in range(len(effect))]
xname2 = ['prob(x1>x2) %s' % ii for ii in xname]
title = "Probability sample 1 is stochastically larger"
from statsmodels.iolib.summary import summary_params
summ = summary_params((self, effect, sd, statistic, pvalues, ci),
yname=yname,
xname=xname2,
use_t=use_t,
title=title,
alpha=alpha)
return summ
def make_table(self, mask, title, strip_end=True):
res = (self, self.params[mask], self.bse[mask],
self.zvalues[mask], self.pvalues[mask],
self.conf_int(alpha)[mask])
param_names = []
for name in np.array(self.data.param_names)[mask].tolist():
if strip_end:
param_name = '.'.join(name.split('.')[:-1])
else:
param_name = name
if name in self.fixed_params:
param_name = '%s (fixed)' % param_name
param_names.append(param_name)
return summary_params(res, yname=None, xname=param_names,
alpha=alpha, use_t=False, title=title)
def summary(self, alpha=.05):
"""
Returns a summary table for marginal effects
Parameters
----------
alpha : float
Number between 0 and 1. The confidence intervals have the
probability 1-alpha.
Returns
-------
Summary : SummaryTable
A SummaryTable instance
"""
_check_at_is_all(self.margeff_options)
results = self.results
model = results.model
title = model.__class__.__name__ + " Marginal Effects"
method = self.margeff_options['method']
top_left = [
('Dep. Variable:', [model.endog_names]),
('Method:', [method]),
('At:', [self.margeff_options['at']]),
]
from statsmodels.iolib.summary import (Summary, summary_params,
table_extend)
exog_names = model.exog_names[:] # copy
smry = Summary()
# sigh, we really need to hold on to this in _data...
_, const_idx = _get_const_index(model.exog)
if const_idx is not None:
exog_names.pop(const_idx[0])
J = int(getattr(model, "J", 1))
if J > 1:
yname, yname_list = results._get_endog_name(model.endog_names,
None,
all=True)
else:
yname = model.endog_names
yname_list = [yname]
smry.add_table_2cols(self,
gleft=top_left,
gright=[],
yname=yname,
xname=exog_names,
title=title)
#NOTE: add_table_params is not general enough yet for margeff
# could use a refactor with getattr instead of hard-coded params
# tvalues etc.
table = []
conf_int = self.conf_int(alpha)
margeff = self.margeff
margeff_se = self.margeff_se
tvalues = self.tvalues
pvalues = self.pvalues
if J > 1:
for eq in range(J):
restup = (results, margeff[:, eq], margeff_se[:, eq],
tvalues[:, eq], pvalues[:, eq], conf_int[:, :, eq])
tble = summary_params(restup,
yname=yname_list[eq],
xname=exog_names,
alpha=alpha,
use_t=False,
skip_header=True)
tble.title = yname_list[eq]
# overwrite coef with method name
header = [
'', _transform_names[method], 'std err', 'z', 'P>|z|',
'[' + str(alpha / 2),
str(1 - alpha / 2) + ']'
]
tble.insert_header_row(0, header)
#from IPython.core.debugger import Pdb; Pdb().set_trace()
table.append(tble)
table = table_extend(table, keep_headers=True)
else:
restup = (results, margeff, margeff_se, tvalues, pvalues, conf_int)
table = summary_params(restup,
yname=yname,
xname=exog_names,
alpha=alpha,
use_t=False,
skip_header=True)
header = [
'', _transform_names[method], 'std err', 'z', 'P>|z|',
'[' + str(alpha / 2),
str(1 - alpha / 2) + ']'
]
table.insert_header_row(0, header)
smry.tables.append(table)
return smry
def summary(self, alpha=.05):
"""
Returns a summary table for marginal effects
Parameters
----------
alpha : float
Number between 0 and 1. The confidence intervals have the
probability 1-alpha.
Returns
-------
Summary : SummaryTable
A SummaryTable instance
"""
_check_at_is_all(self.margeff_options)
results = self.results
model = results.model
title = model.__class__.__name__ + " Marginal Effects"
method = self.margeff_options['method']
top_left = [('Dep. Variable:', [model.endog_names]),
('Method:', [method]),
('At:', [self.margeff_options['at']]),]
from statsmodels.iolib.summary import (Summary, summary_params,
table_extend)
exog_names = model.exog_names[:] # copy
smry = Summary()
# sigh, we really need to hold on to this in _data...
_, const_idx = _get_const_index(model.exog)
if const_idx is not None:
exog_names.pop(const_idx)
J = int(getattr(model, "J", 1))
if J > 1:
yname, yname_list = results._get_endog_name(model.endog_names,
None, all=True)
else:
yname = model.endog_names
yname_list = [yname]
smry.add_table_2cols(self, gleft=top_left, gright=[],
yname=yname, xname=exog_names, title=title)
#NOTE: add_table_params is not general enough yet for margeff
# could use a refactor with getattr instead of hard-coded params
# tvalues etc.
table = []
conf_int = self.conf_int(alpha)
margeff = self.margeff
margeff_se = self.margeff_se
tvalues = self.tvalues
pvalues = self.pvalues
if J > 1:
for eq in range(J):
restup = (results, margeff[:,eq], margeff_se[:,eq],
tvalues[:,eq], pvalues[:,eq], conf_int[:,:,eq])
tble = summary_params(restup, yname=yname_list[eq],
xname=exog_names, alpha=alpha, use_t=False,
skip_header=True)
tble.title = yname_list[eq]
# overwrite coef with method name
header = ['', _transform_names[method], 'std err', 'z',
'P>|z|', '[%3.1f%% Conf. Int.]' % (100-alpha*100)]
tble.insert_header_row(0, header)
#from IPython.core.debugger import Pdb; Pdb().set_trace()
table.append(tble)
table = table_extend(table, keep_headers=True)
else:
restup = (results, margeff, margeff_se, tvalues, pvalues, conf_int)
table = summary_params(restup, yname=yname, xname=exog_names,
alpha=alpha, use_t=False, skip_header=True)
header = ['', _transform_names[method], 'std err', 'z',
'P>|z|', '[%3.1f%% Conf. Int.]' % (100-alpha*100)]
table.insert_header_row(0, header)
smry.tables.append(table)
return smry
def get_parameter_summary(self):
print(summary_params(self.logit_result_linear).data)
