def __init__(self, d=None, name=''):
if d == None:
d = {}
test_obj_instance(d, (dict))
self.d = d
self.name = name
def make_transform(distro, transform_type='', complement=False):
test_obj_instance(distro, DictWrapper)
xs, ps = distro.sort_zip()
scale = dict(xscale='linear', yscale='linear')
if transform_type == 'exponential':
complement = True
scale['yscale'] = 'log'
if transform_type == 'pareto':
complement = True
scale['yscale'] = 'log'
scale['xscale'] = 'log'
if complement:
ps = [1.0-p for p in ps]
if transform_type == 'weibull':
xs.pop()
ps.pop()
ps = [-math.log(1.0-p) for p in ps]
scale['xscale'] = 'log'
scale['yscale'] = 'log'
if transform_type == 'gumbel':
xs.pop(0)
ps.pop(0)
ps = [-math.log(p) for p in ps]
scale['yscale'] = 'log'
return distro.__class__(dict(zip(xs,ps)), distro.name), scale
def make_mixture_pmfs(pmfs, name='mix'):
test_obj_instance(pmfs, dict)
mix = Pmf(name=name)
for pmf, prob in pmfs.iteritems():
for x, p in pmf:
mix.increment(x, p * prob)
pmf.normalise()
return mix
def __init__(self, prior, sample, likelihood):
test_obj_instance(prior, Pmf)
test_obj_instance(sample, list)
test_obj_subclass(likelihood, Likelihood)
self.prior = prior
self.sample = sample
self.likelihood = likelihood
self.posterior = deepcopy(prior)
self.posterior.name = 'posterior'
self._estimate_posterior()
def __init__(self, prior, sample, likelihood):
test_obj_instance(prior, Pmf)
test_obj_instance(sample, list)
test_obj_subclass(likelihood, Likelihood)
self.prior = prior
self.sample = sample
self.likelihood = likelihood
self.posterior = deepcopy(prior)
self.posterior.name = 'posterior'
self._estimate_posterior()
def make_pmf_from_hist(hist, name=None):
if name is None:
name = hist.name
test_obj_instance(hist, Histogram)
# make a copy of the dictionary
d = dict(hist.get_dict())
pmf = Pmf(d, name)
pmf.normalise()
return pmf
def relative_mean_difference(pmf, mu=None):
test_obj_instance(pmf, Pmf)
if mu is None:
mu = mean(pmf)
diff = Pmf()
for v1, p1 in pmf.items():
for v2, p2 in pmf.items():
diff.increment(abs(v1 - v2), p1 * p2)
return mean(diff) / mu
def make_pmf_from_cdf(cdf, name=None):
if name is None:
name = cdf.name
test_obj_instance(cdf, Cdf)
pmf = Pmf(name=name)
prev = 0.0
for val, prob in cdf:
pmf.increment(val, prob-prev)
prev = prob
pmf.normalise()
return pmf
def __init__(self, null_hypo_results, alternative_hypo_result, bayesian_result_suite):
test_obj_instance(null_hypo_results, HypoTestResult)
test_obj_instance(alternative_hypo_result, HypoTestResult)
test_obj_instance(bayesian_result_suite, BayesianResultSuite)
self.null_hypo_results = null_hypo_results
self.alternative_hypo_result = alternative_hypo_result
self.bayesian_result_suite = bayesian_result_suite
def __init__(self, median_bayesian_result, plus_sigma_bayesian_result, minus_sigma_bayesian_result):
test_obj_instance(median_bayesian_result, BayesianResult)
test_obj_instance(plus_sigma_bayesian_result, BayesianResult)
test_obj_instance(minus_sigma_bayesian_result, BayesianResult)
self.median_bayesian_result = median_bayesian_result
self.plus_sigma_bayesian_result = plus_sigma_bayesian_result
self.minus_sigma_bayesian_result = minus_sigma_bayesian_result
def __init__(self, null_hypo_results, alternative_hypo_result,
bayesian_result_suite):
test_obj_instance(null_hypo_results, HypoTestResult)
test_obj_instance(alternative_hypo_result, HypoTestResult)
test_obj_instance(bayesian_result_suite, BayesianResultSuite)
self.null_hypo_results = null_hypo_results
self.alternative_hypo_result = alternative_hypo_result
self.bayesian_result_suite = bayesian_result_suite
def __init__(self, median_bayesian_result, plus_sigma_bayesian_result,
minus_sigma_bayesian_result):
test_obj_instance(median_bayesian_result, BayesianResult)
test_obj_instance(plus_sigma_bayesian_result, BayesianResult)
test_obj_instance(minus_sigma_bayesian_result, BayesianResult)
self.median_bayesian_result = median_bayesian_result
self.plus_sigma_bayesian_result = plus_sigma_bayesian_result
self.minus_sigma_bayesian_result = minus_sigma_bayesian_result
def _validate_objects(self):
test_obj_instance(self.sample_one, list)
test_obj_instance(self.sample_two, list)
test_obj_instance(self.population, list)
test_obj_instance(self.options, dict)
def __init__(self, results):
test_obj_instance(results, dict)
self.results = results
def make_pmf_from_list(lst, name=''):
test_obj_instance(lst, list)
hist = make_hist_from_list(lst, name)
return make_pmf_from_hist(hist)
def _validate_input_objects(self, observed_sample, expected_sample, prob_functions):
test_obj_instance(observed_sample, Pmf)
test_obj_instance(expected_sample, Pmf)
def plot_histograms(distros, **options):
test_obj_instance(distros, list)
for distro in distros:
plot_histogram(distro, **options)
def make_cdf_from_dict(dct, name=''):
test_obj_instance(dct, dict)
return make_cdf_from_items(dct.items(), name)
def make_hist_from_list(lst, name=''):
test_obj_instance(lst, list)
hist = Histogram(name=name)
[hist.increment(x) for x in lst]
return hist
def make_pmf_from_dict(dct, name=''):
test_obj_instance(dct, dict)
pmf = Pmf(dct, name)
pmf.normalise()
return pmf
def __init__(self, results):
test_obj_instance(results, dict)
self.results = results
def plot_discrete_distributions(distros, **options):
test_obj_instance(distros, list)
for distro in distros:
plot_discrete_distribution(distro, **options)
def get_probabilities(self, pmf):
test_obj_instance(pmf, Pmf)
cdf = make_cdf_from_pmf(pmf)
return cdf.get_prob(self.high) - cdf.get_prob(self.low)
def _validate_objects(self):
test_obj_instance(self.sample_one, list)
test_obj_instance(self.sample_two, list)
test_obj_instance(self.population, list)
test_obj_instance(self.options, dict)
def __init__(self, sample):
test_obj_instance(sample, list)
self.__sample = sample
self.__mean = mean(sample)
self.__median = median(sample)
def make_hist_from_dict(dct, name=''):
test_obj_instance(dct, dict)
return Histogram(dct, name)