def test_chisquare_power():
from .results.results_power import pwr_chisquare
for case in itervalues(pwr_chisquare):
power = chisquare_power(
case.w, case.N, case.df + 1, alpha=case.sig_level)
assert_almost_equal(
power, case.power, decimal=6, err_msg=repr(vars(case)))
def power(self, effect_size, nobs, n_bins, alpha, ddof=0):
#alternative='two-sided'):
'''Calculate the power of a chisquare test for one sample
Only two-sided alternative is implemented
Parameters
----------
effect_size : float
standardized effect size, according to Cohen's definition.
see :func:`statsmodels.stats.gof.chisquare_effectsize`
nobs : int or float
sample size, number of observations.
alpha : float in interval (0,1)
significance level, e.g. 0.05, is the probability of a type I
error, that is wrong rejections if the Null Hypothesis is true.
n_bins : int
number of bins or cells in the distribution.
Returns
-------
power : float
Power of the test, e.g. 0.8, is one minus the probability of a
type II error. Power is the probability that the test correctly
rejects the Null Hypothesis if the Alternative Hypothesis is true.
'''
from statsmodels.stats.gof import chisquare_power
return chisquare_power(effect_size, nobs, n_bins, alpha, ddof=0)
def power(self,
effect_size,
nobs,
alpha,
n_bins,
ddof=0): #alternative='two-sided'):
'''Calculate the power of a chisquare test for one sample
Only two-sided alternative is implemented
Parameters
----------
effect_size : float
standardized effect size, according to Cohen's definition.
see :func:`statsmodels.stats.gof.chisquare_effectsize`
nobs : int or float
sample size, number of observations.
alpha : float in interval (0,1)
significance level, e.g. 0.05, is the probability of a type I
error, that is wrong rejections if the Null Hypothesis is true.
n_bins : int
number of bins or cells in the distribution.
Returns
-------
power : float
Power of the test, e.g. 0.8, is one minus the probability of a
type II error. Power is the probability that the test correctly
rejects the Null Hypothesis if the Alternative Hypothesis is true.
'''
from statsmodels.stats.gof import chisquare_power
return chisquare_power(effect_size, nobs, n_bins, alpha, ddof=0)
def power(self,
effect_size,
nobs1,
alpha,
ratio=1,
alternative='two-sided'):
'''Calculate the power of a chisquare for two independent sample
Parameters
----------
effect_size : float
standardize effect size, difference between the two means divided
by the standard deviation. effect size has to be positive.
nobs1 : int or float
number of observations of sample 1. The number of observations of
sample two is ratio times the size of sample 1,
i.e. ``nobs2 = nobs1 * ratio``
alpha : float in interval (0,1)
significance level, e.g. 0.05, is the probability of a type I
error, that is wrong rejections if the Null Hypothesis is true.
ratio : float
ratio of the number of observations in sample 2 relative to
sample 1. see description of nobs1
The default for ratio is 1; to solve for ration given the other
arguments it has to be explicitely set to None.
alternative : str, 'two-sided' (default) or 'one-sided'
extra argument to choose whether the power is calculated for a
two-sided (default) or one sided test.
'one-sided' assumes we are in the relevant tail.
Returns
-------
power : float
Power of the test, e.g. 0.8, is one minus the probability of a
type II error. Power is the probability that the test correctly
rejects the Null Hypothesis if the Alternative Hypothesis is true.
'''
from statsmodels.stats.gof import chisquare_power
nobs2 = nobs1 * ratio
#equivalent to nobs = n1*n2/(n1+n2)=n1*ratio/(1+ratio)
nobs = 1. / (1. / nobs1 + 1. / nobs2)
return chisquare_power(effect_size, nobs, alpha)
def power(self, effect_size, nobs1, alpha, ratio=1,
alternative='two-sided'):
'''Calculate the power of a chisquare for two independent sample
Parameters
----------
effect_size : float
standardize effect size, difference between the two means divided
by the standard deviation. effect size has to be positive.
nobs1 : int or float
number of observations of sample 1. The number of observations of
sample two is ratio times the size of sample 1,
i.e. ``nobs2 = nobs1 * ratio``
alpha : float in interval (0,1)
significance level, e.g. 0.05, is the probability of a type I
error, that is wrong rejections if the Null Hypothesis is true.
ratio : float
ratio of the number of observations in sample 2 relative to
sample 1. see description of nobs1
The default for ratio is 1; to solve for ration given the other
arguments it has to be explicitely set to None.
alternative : string, 'two-sided' (default) or 'one-sided'
extra argument to choose whether the power is calculated for a
two-sided (default) or one sided test.
'one-sided' assumes we are in the relevant tail.
Returns
-------
power : float
Power of the test, e.g. 0.8, is one minus the probability of a
type II error. Power is the probability that the test correctly
rejects the Null Hypothesis if the Alternative Hypothesis is true.
'''
from statsmodels.stats.gof import chisquare_power
nobs2 = nobs1*ratio
#equivalent to nobs = n1*n2/(n1+n2)=n1*ratio/(1+ratio)
nobs = 1./ (1. / nobs1 + 1. / nobs2)
return chisquare_power(effect_size, nobs, alpha)
def log_results(log, result, source):
"""
Log the fitting results.
Notes
-----
The resulting mixture parameters are stored into a 2d array with rows
[location in degrees (mu), shape (kappa), probability].
"""
sparams = result.model.get_summary_params(result.full_params)[:, [1, 0, 2]]
sparams[:, 0] = tr.transform_pi_deg(tr.fix_range(sparams[:, 0]),
neg_shift=source.neg_shift)
converged = result.mle_retvals['converged']
fit_criteria = [-result.llf, result.aic, result.bic]
print 'llf / nobs:', fit_criteria[0] / result.model.endog.shape[0]
chisquare = result.gof_chisquare()
# Chisquare test with effect size.
alpha = 0.05 # Significance level.
data = source.source_data.data
n_obs = data[:, 1].sum()
rad_diff = data[1, 0] - data[0, 0]
pdf = result.model.pdf_mix(result.full_params, data[:, 0])
probs = pdf * rad_diff * n_obs
effect_size = gof.chisquare_effectsize(data[:, 1], probs)
chi2 = gof.chisquare(data[:, 1], probs, value=effect_size)
power = gof.chisquare_power(effect_size, n_obs,
data.shape[0], alpha=alpha)
chisquare_all = list(chisquare) + [n_obs, effect_size] \
+ list(chi2) + [power]
log.write_row(source.current.dir_base, source.current.base_names,
chisquare_all, sparams, converged, fit_criteria)
def chisquare_test_scores(self, E1_score_list, E2_score_list):
obs = np.array([E1_score_list, E2_score_list])
print(obs)
#obs = np.array([female_list]).T
chi2_stat, p_val, dof, ex = chi2_contingency(obs, correction=False)
#compute effect size
effect_size = sms.chisquare_effectsize(E1_score_list, E2_score_list)
number_observations = len(E1_score_list)
#compute power
from statsmodels.stats.gof import chisquare_power
power = chisquare_power(effect_size,
nobs=number_observations,
n_bins=dof + 1,
alpha=0.05)
#smp.GofChisquarePower.solve_power(effect_size, nobs=number_observations, n_bins=degrees_freedom+1, alpha=0.05)
print("Chi-square Stat: " + str(chi2_stat))
print(" degrees of freedom: " + str(dof))
print(" p-value: " + str(p_val))
print(" contingency table: " + str(ex))
print(" effect size:" + str(effect_size))
print(" power:" + str(power))
def log_results(log, result, source):
"""
Log the fitting results.
Notes
-----
The resulting mixture parameters are stored into a 2d array with rows
[location in degrees (mu), shape (kappa), probability].
"""
sparams = result.model.get_summary_params(result.full_params)[:, [1, 0, 2]]
sparams[:, 0] = tr.transform_pi_deg(tr.fix_range(sparams[:, 0]),
neg_shift=source.neg_shift)
converged = result.mle_retvals['converged']
fit_criteria = [-result.llf, result.aic, result.bic]
print 'llf / nobs:', fit_criteria[0] / result.model.endog.shape[0]
chisquare = result.gof_chisquare()
# Chisquare test with effect size.
alpha = 0.05 # Significance level.
data = source.source_data.data
n_obs = data[:, 1].sum()
rad_diff = data[1, 0] - data[0, 0]
pdf = result.model.pdf_mix(result.full_params, data[:, 0])
probs = pdf * rad_diff * n_obs
effect_size = gof.chisquare_effectsize(data[:, 1], probs)
chi2 = gof.chisquare(data[:, 1], probs, value=effect_size)
power = gof.chisquare_power(effect_size, n_obs, data.shape[0], alpha=alpha)
chisquare_all = list(chisquare) + [n_obs, effect_size] \
+ list(chi2) + [power]
log.write_row(source.current.dir_base, source.current.base_names,
chisquare_all, sparams, converged, fit_criteria)
print(stats.ncx2.sf(chisq, n_bins, 0.001 * nobs))
print(stats.ncx2.sf(chisq, n_bins, d_delta * nobs))
print(chisquare(freq, nobs*probs_d, value=np.sqrt(d_delta)))
print(chisquare(freq, nobs*probs_d, value=np.sqrt(chisq / nobs)))
print()
assert_almost_equal(stats.chi2.sf(d_delta * nobs, n_bins - 1),
chisquare(freq, nobs*probs_d)[1], decimal=13)
crit = stats.chi2.isf(0.05, n_bins - 1)
power = stats.ncx2.sf(crit, n_bins-1, 0.001**2 * nobs)
#> library(pwr)
#> tr = pwr.chisq.test(w =0.001, N =30000 , df = 5-1, sig.level = 0.05, power = NULL)
assert_almost_equal(power, 0.05147563, decimal=7)
effect_size = 0.001
power = chisquare_power(effect_size, nobs, n_bins, alpha=0.05)
assert_almost_equal(power, 0.05147563, decimal=7)
print(chisquare(freq, nobs*probs, value=0, ddof=0))
d_null_alt = ((probs - probs_d)**2 / probs).sum()
print(chisquare(freq, nobs*probs, value=np.sqrt(d_null_alt), ddof=0))
#Monte Carlo to check correct size and power of test
d_delta_r = chisquare_effectsize(probs, probs_d)
n_rep = 10000
nobs = 3000
res_boots = np.zeros((n_rep, 6))
for i in range(n_rep):
rvs = np.argmax(np.random.rand(nobs,1) < probs_cs, 1)
freq = np.bincount(rvs)