def test_weightstats_3(self):
x1_2d, x2_2d = self.x1_2d, self.x2_2d
w1, w2 = self.w1, self.w2
d1w_2d = DescrStatsW(x1_2d, weights=w1)
d2w_2d = DescrStatsW(x2_2d, weights=w2)
x1r_2d = d1w_2d.asrepeats()
x2r_2d = d2w_2d.asrepeats()
assert_almost_equal(x2r_2d.mean(0), d2w_2d.mean, 14)
assert_almost_equal(x2r_2d.var(0), d2w_2d.var, 14)
assert_almost_equal(x2r_2d.std(0), d2w_2d.std, 14)
assert_almost_equal(np.cov(x2r_2d.T, bias=1), d2w_2d.cov, 14)
assert_almost_equal(np.corrcoef(x2r_2d.T), d2w_2d.corrcoef, 14)
# print d1w_2d.ttest_mean(3)
# #scipy.stats.ttest is also vectorized
# print stats.ttest_1samp(x1r_2d, 3)
t, p, d = d1w_2d.ttest_mean(3)
assert_almost_equal([t, p], stats.ttest_1samp(x1r_2d, 3), 11)
# print [stats.ttest_1samp(xi, 3) for xi in x1r_2d.T]
cm = CompareMeans(d1w_2d, d2w_2d)
ressm = cm.ttest_ind()
resss = stats.ttest_ind(x1r_2d, x2r_2d)
assert_almost_equal(ressm[:2], resss, 14)
def test_weightstats_3(self):
x1_2d, x2_2d = self.x1_2d, self.x2_2d
w1, w2 = self.w1, self.w2
d1w_2d = DescrStatsW(x1_2d, weights=w1)
d2w_2d = DescrStatsW(x2_2d, weights=w2)
x1r_2d = d1w_2d.asrepeats()
x2r_2d = d2w_2d.asrepeats()
assert_almost_equal(x2r_2d.mean(0), d2w_2d.mean, 14)
assert_almost_equal(x2r_2d.var(0), d2w_2d.var, 14)
assert_almost_equal(x2r_2d.std(0), d2w_2d.std, 14)
assert_almost_equal(np.cov(x2r_2d.T, bias=1), d2w_2d.cov, 14)
assert_almost_equal(np.corrcoef(x2r_2d.T), d2w_2d.corrcoef, 14)
# print d1w_2d.ttest_mean(3)
# #scipy.stats.ttest is also vectorized
# print stats.ttest_1samp(x1r_2d, 3)
t, p, d = d1w_2d.ttest_mean(3)
assert_almost_equal([t, p], stats.ttest_1samp(x1r_2d, 3), 11)
#print [stats.ttest_1samp(xi, 3) for xi in x1r_2d.T]
cm = CompareMeans(d1w_2d, d2w_2d)
ressm = cm.ttest_ind()
resss = stats.ttest_ind(x1r_2d, x2r_2d)
assert_almost_equal(ressm[:2], resss, 14)
def test_confint(testdata):
result = confidence_interval(testdata, control_label='A')
c_means1 = CompareMeans(DescrStatsW(testdata['kpi1']['B']),
DescrStatsW(testdata['kpi1']['A']))
c_means2 = CompareMeans(DescrStatsW(testdata['kpi2']['B']),
DescrStatsW(testdata['kpi2']['A']))
expected1 = c_means1.tconfint_diff()
expected2 = c_means2.zconfint_diff()
assert result['B']['kpi1'] == expected1
assert result['B']['kpi2'] == expected2
def test_comparemeans_convenient_interface(self):
x1_2d, x2_2d = self.x1_2d, self.x2_2d
d1 = DescrStatsW(x1_2d)
d2 = DescrStatsW(x2_2d)
cm1 = CompareMeans(d1, d2)
# smoke test for summary
from statsmodels.iolib.table import SimpleTable
for use_t in [True, False]:
for usevar in ['pooled', 'unequal']:
smry = cm1.summary(use_t=use_t, usevar=usevar)
assert_(isinstance(smry, SimpleTable))
# test for from_data method
cm2 = CompareMeans.from_data(x1_2d, x2_2d)
assert_(str(cm1.summary()) == str(cm2.summary()))
def setup_class(cls):
cls.x1 = np.array(
[7.8, 6.6, 6.5, 7.4, 7.3, 7., 6.4, 7.1, 6.7, 7.6, 6.8])
cls.x2 = np.array([4.5, 5.4, 6.1, 6.1, 5.4, 5., 4.1, 5.5])
cls.d1 = DescrStatsW(cls.x1)
cls.d2 = DescrStatsW(cls.x2)
cls.cm = CompareMeans(cls.d1, cls.d2)
def test_ztest_ztost():
# compare weightstats with separately tested proportion ztest ztost
import statsmodels.stats.proportion as smprop
x1 = [0, 1]
w1 = [5, 15]
res2 = smprop.proportions_ztest(15, 20., value=0.5)
d1 = DescrStatsW(x1, w1)
res1 = d1.ztest_mean(0.5)
assert_allclose(res1, res2, rtol=0.03, atol=0.003)
d2 = DescrStatsW(x1, np.array(w1) * 21. / 20)
res1 = d2.ztest_mean(0.5)
assert_almost_equal(res1, res2, decimal=12)
res1 = d2.ztost_mean(0.4, 0.6)
res2 = smprop.proportions_ztost(15, 20., 0.4, 0.6)
assert_almost_equal(res1[0], res2[0], decimal=12)
x2 = [0, 1]
w2 = [10, 10]
#d2 = DescrStatsW(x1, np.array(w1)*21./20)
d2 = DescrStatsW(x2, w2)
res1 = ztest(d1.asrepeats(), d2.asrepeats())
res2 = smprop.proportions_chisquare(np.asarray([15, 10]),
np.asarray([20., 20]))
#TODO: check this is this difference expected?, see test_proportion
assert_allclose(res1[1], res2[1], rtol=0.03)
res1a = CompareMeans(d1, d2).ztest_ind()
assert_allclose(res1a[1], res2[1], rtol=0.03)
assert_almost_equal(res1a, res1, decimal=12)
def test_comparemeans_convenient_interface(self):
x1_2d, x2_2d = self.x1_2d, self.x2_2d
d1 = DescrStatsW(x1_2d)
d2 = DescrStatsW(x2_2d)
cm1 = CompareMeans(d1, d2)
# smoke test for summary
from statsmodels.iolib.table import SimpleTable
for use_t in [True, False]:
for usevar in ['pooled', 'unequal']:
smry = cm1.summary(use_t=use_t, usevar=usevar)
assert_(isinstance(smry, SimpleTable))
# test for from_data method
cm2 = CompareMeans.from_data(x1_2d, x2_2d)
assert_(str(cm1.summary()) == str(cm2.summary()))
def testcomparison(df, smp1_cols, smp2_cols, test='ttest'):
if len(smp1_cols) == 0 or len(smp2_cols) == 0:
logging.warning("data not exist for comparison")
else:
col_stat = 'stat %s' % test
col_pval = 'pval %s' % test
df.loc[:, col_stat] = np.nan
df.loc[:, col_pval] = np.nan
for i in df.index:
X = DescrStatsW(df.loc[i, smp1_cols].as_matrix())
Y = DescrStatsW(df.loc[i, smp2_cols].as_matrix())
if test == 'ttest':
df.loc[i, col_stat], df.loc[i, col_pval], tmp = CompareMeans(
X, Y).ttest_ind()
if test == 'ztest':
df.loc[i, col_stat], df.loc[i, col_pval] = CompareMeans(
X, Y).ztest_ind()
return df
def mean_diff_confint_ind(sample1, sample2, alpha=0.05):
"""Доверительный интервал разности средних для двух независимых выборок
Parameters
----------
sample1 : array_like
Первая выборка
sample2 : array_like
Вторая выборка
alpha : float in (0, 1)
Уровень доверия, рассчитывается как ``1-alpha``
Returns
-------
lower, upper : floats
Левая и правая граница доверительного интервала
"""
cm = CompareMeans(DescrStatsW(sample1), DescrStatsW(sample2))
return cm.tconfint_diff(alpha=alpha)
def test_means(self, population_A, population_B, sample_size_A,
sample_size_B, variable, SEED):
# gets a sample for each popuplation
sample_A = population_A.sample(n=sample_size_A,
random_state=SEED)[variable]
sample_B = population_B.sample(n=sample_size_B,
random_state=SEED)[variable]
# prepares the tests
test_pop_A = DescrStatsW(sample_A)
test_pop_B = DescrStatsW(sample_B)
# makes the comparison
test_compare = CompareMeans(test_pop_A, test_pop_B)
# tests
z, p_value = test_compare.ztest_ind(alternative='larger', value=0)
return p_value
def mean_method(df, target_col):
df_t0 = df.loc[(df['treatment'] == 0) & (df['in_delta'] == 1)]
df_t1 = df.loc[(df['treatment'] == 1) & (df['in_delta'] == 1)]
df_t0 = df_t0[target_col]
df_t1 = df_t1[target_col]
mean_0 = df_t0.mean()
mean_1 = df_t1.mean()
print(f"{'=' * 10}Mean Method Results{'=' * 10}")
effect = mean_1 - mean_0
print(f"Treatment effect on {target_col} is {mean_1 - mean_0}")
cm = CompareMeans.from_data(df_t1.values, df_t0.values)
result = cm.ttest_ind(alternative='two-sided')
print(f"Two-sided T test: CI={np.round(cm.tconfint_diff(), 3)} pvalue={round(result[1], 3)}")
return effect
def mean_diff_confint_ind(sample1, sample2):
cm = CompareMeans(DescrStatsW(sample1), DescrStatsW(sample2))
return cm.tconfint_diff()
def test_ttest_2sample(self):
x1, x2 = self.x1, self.x2
x1r, x2r = self.x1r, self.x2r
w1, w2 = self.w1, self.w2
#Note: stats.ttest_ind handles 2d/nd arguments
res_sp = stats.ttest_ind(x1r, x2r)
assert_almost_equal(
ttest_ind(x1, x2, weights=(w1, w2))[:2], res_sp, 14)
#check correct ttest independent of user ddof
cm = CompareMeans(DescrStatsW(x1, weights=w1, ddof=0),
DescrStatsW(x2, weights=w2, ddof=1))
assert_almost_equal(cm.ttest_ind()[:2], res_sp, 14)
cm = CompareMeans(DescrStatsW(x1, weights=w1, ddof=1),
DescrStatsW(x2, weights=w2, ddof=2))
assert_almost_equal(cm.ttest_ind()[:2], res_sp, 14)
cm0 = CompareMeans(DescrStatsW(x1, weights=w1, ddof=0),
DescrStatsW(x2, weights=w2, ddof=0))
cm1 = CompareMeans(DescrStatsW(x1, weights=w1, ddof=0),
DescrStatsW(x2, weights=w2, ddof=1))
cm2 = CompareMeans(DescrStatsW(x1, weights=w1, ddof=1),
DescrStatsW(x2, weights=w2, ddof=2))
res0 = cm0.ttest_ind(usevar='unequal')
res1 = cm1.ttest_ind(usevar='unequal')
res2 = cm2.ttest_ind(usevar='unequal')
assert_almost_equal(res1, res0, 14)
assert_almost_equal(res2, res0, 14)
#check confint independent of user ddof
res0 = cm0.tconfint_diff(usevar='pooled')
res1 = cm1.tconfint_diff(usevar='pooled')
res2 = cm2.tconfint_diff(usevar='pooled')
assert_almost_equal(res1, res0, 14)
assert_almost_equal(res2, res0, 14)
res0 = cm0.tconfint_diff(usevar='unequal')
res1 = cm1.tconfint_diff(usevar='unequal')
res2 = cm2.tconfint_diff(usevar='unequal')
assert_almost_equal(res1, res0, 14)
assert_almost_equal(res2, res0, 14)
def fn(control, test):
c_means = CompareMeans(DescrStatsW(test), DescrStatsW(control))
if _is_proportion(control, test):
return c_means.zconfint_diff()
else:
return c_means.tconfint_diff()
def test_ttest_2sample(self):
x1, x2 = self.x1, self.x2
x1r, x2r = self.x1r, self.x2r
w1, w2 = self.w1, self.w2
# Note: stats.ttest_ind handles 2d/nd arguments
res_sp = stats.ttest_ind(x1r, x2r)
assert_almost_equal(ttest_ind(x1, x2, weights=(w1, w2))[:2],
res_sp, 14)
# check correct ttest independent of user ddof
cm = CompareMeans(DescrStatsW(x1, weights=w1, ddof=0),
DescrStatsW(x2, weights=w2, ddof=1))
assert_almost_equal(cm.ttest_ind()[:2], res_sp, 14)
cm = CompareMeans(DescrStatsW(x1, weights=w1, ddof=1),
DescrStatsW(x2, weights=w2, ddof=2))
assert_almost_equal(cm.ttest_ind()[:2], res_sp, 14)
cm0 = CompareMeans(DescrStatsW(x1, weights=w1, ddof=0),
DescrStatsW(x2, weights=w2, ddof=0))
cm1 = CompareMeans(DescrStatsW(x1, weights=w1, ddof=0),
DescrStatsW(x2, weights=w2, ddof=1))
cm2 = CompareMeans(DescrStatsW(x1, weights=w1, ddof=1),
DescrStatsW(x2, weights=w2, ddof=2))
res0 = cm0.ttest_ind(usevar='unequal')
res1 = cm1.ttest_ind(usevar='unequal')
res2 = cm2.ttest_ind(usevar='unequal')
assert_almost_equal(res1, res0, 14)
assert_almost_equal(res2, res0, 14)
# check confint independent of user ddof
res0 = cm0.tconfint_diff(usevar='pooled')
res1 = cm1.tconfint_diff(usevar='pooled')
res2 = cm2.tconfint_diff(usevar='pooled')
assert_almost_equal(res1, res0, 14)
assert_almost_equal(res2, res0, 14)
res0 = cm0.tconfint_diff(usevar='unequal')
res1 = cm1.tconfint_diff(usevar='unequal')
res2 = cm2.tconfint_diff(usevar='unequal')
assert_almost_equal(res1, res0, 14)
assert_almost_equal(res2, res0, 14)
info = max(values, key=itemgetter(2))
a = info[0] - 1
b = info[1] - 1
x = range(1, 10)
n = a + 1
l = "User:"******"User:"******"Pearson:", scipy.stats.pearsonr(R[a], R[b]))
print(CompareMeans(DescrStatsW(R[a]), DescrStatsW(R[b])).summary())
results = sm.OLS(R[b], R[a]).fit()
print(results.summary())
for i in range(0, 9):
if R[a][i] == 0:
if R[b][i] >= 3:
print("product", i + 1, "recommended to user", a + 1)
for i in range(0, 9):
if R[b][i] == 0:
if R[a][i] >= 3:
print("product", i + 1, "recommended to user", b + 1)
# Critério do valor p
# Teste Unicaudal
# Rejeitar H_0 se o valor p\leq\alpha
from statsmodels.stats.weightstats import DescrStatsW, CompareMeans
test_setosa = DescrStatsW(setosa)
test_virginica = DescrStatsW(virginica)
test_A = test_setosa.get_compare(test_virginica)
z, p_valor = test_A.ztest_ind(alternative='larger', value=0)
print('O valor de z é ', z)
print('O p-valor é ', p_valor)
test_B = CompareMeans(test_setosa, test_virginica)
z, p_valor = test_B.ztest_ind(alternative='larger', value=0)
print('O valor de z é ', z)
print('O p-valor é ', p_valor)
p_valor <= significancia
# Aceitamos a hipótese nula.
# Testes não Paramétricos
"""
Teste do Qui-Quadrado (\chi^2)
Também conhecido como teste de adequação ao ajustamento, seu nome se
deve ao fato de utilizar uma variável estatística padronizada, representada pela
32.6, 33.3, 32.2
]
a = 0.05 # 有意水準（デフォルト） = 1 - 信頼係数
alt = 'two-sided' # 両側検定（デフォルト）
# 左片側検定なら'smaller'
# 右片側検定なら'larger'
d = DescrStatsW(np.array(X) - np.array(Y)) # 対標本の場合
d.ttest_mean(alternative=alt)[1] # p値
#> 0.0006415571512322235
d.tconfint_mean(alpha=a, alternative=alt) # 信頼区間
#> (-3.9955246743198867, -1.3644753256801117)
c = CompareMeans(DescrStatsW(X), DescrStatsW(Y)) # 対標本でない場合
ve = 'pooled' # 等分散を仮定する（デフォルト）．仮定しないなら'unequal'．
c.ttest_ind(alternative=alt, usevar=ve)[1] # p値
#> 0.000978530937238609
c.tconfint_diff(alpha=a, alternative=alt, usevar=ve) # 信頼区間
#> (-4.170905570517185, -1.1890944294828283)
### 4.4.4 独立性の検定（カイ2乗検定）
import pandas as pd
my_url = ('https://raw.githubusercontent.com/taroyabuki'
'/fromzero/master/data/smoker.csv')
my_data = pd.read_csv(my_url)
