def setup(self):
self.data = pd.DataFrame([
{
'group': "1",
'count': 5000,
'sum': 10021.0,
'sum_of_squares': 25142.0,
'avg': 2.004210,
'var': 1.0116668
},
{
'group': "2",
'count': 5000,
'sum': 9892.0,
'sum_of_squares': 24510.0,
'avg': 1.978424,
'var': 0.9881132
},
])
self.test = spotify_confidence.StudentsTTest(
self.data,
numerator_column='sum',
numerator_sum_squares_column='sum_of_squares',
denominator_column='count',
categorical_group_columns='group',
interval_size=0.99)
def setup(self):
np.random.seed(123)
self.data = pd.DataFrame({
'variation_name':
['test', 'control', 'test2', 'test', 'control', 'test2'],
'nr_of_items': [1969, 312, 2955, 195, 24, 330],
'nr_of_items_sumsq': [5767, 984, 8771, 553, 80, 1010],
'users': [1009, 104, 1502, 100, 10, 150],
'country': [
'us',
'us',
'us',
'gb',
'gb',
'gb',
]
})
self.test = spotify_confidence.StudentsTTest(
self.data,
numerator_column='nr_of_items',
numerator_sum_squares_column='nr_of_items_sumsq',
denominator_column='users',
categorical_group_columns=['country', 'variation_name'],
interval_size=0.95)
def setup(self):
self.data = pd.DataFrame({
'variation_name': [
'test', 'control', 'test2', 'test', 'control', 'test2', 'test',
'control', 'test2', 'test', 'control', 'test2', 'test',
'control', 'test2'
],
'nr_of_items':
[500, 8, 100, 510, 8, 100, 520, 9, 104, 530, 7, 100, 530, 8, 103],
'nr_of_items_sumsq': [
2500, 12, 150, 2510, 13, 140, 2520, 14, 154, 2530, 15, 160,
2530, 16, 103
],
'users': [
1010, 22, 150, 1000, 20, 153, 1030, 23, 154, 1000, 20, 150,
1040, 21, 155
],
'days_since_reg': [1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5],
})
self.test = spotify_confidence.StudentsTTest(
self.data,
numerator_column='nr_of_items',
numerator_sum_squares_column='nr_of_items_sumsq',
denominator_column='users',
categorical_group_columns='variation_name',
ordinal_group_column='days_since_reg',
interval_size=0.95)
def test_p_value_is_symmetric_and_correct(self):
df = pd.DataFrame({
'group_name': ['Control', 'Test'],
'users': [48351, 50571],
'nr_of_items': [1.438602e+06, 1.521974e+06],
'nr_of_items_sumsq': [7.330581e+07, 7.862121e+07]
})
ttest = spotify_confidence.StudentsTTest(
data_frame=df,
denominator_column='users',
numerator_column='nr_of_items',
numerator_sum_squares_column='nr_of_items_sumsq',
categorical_group_columns='group_name')
diff_summary_1 = ttest.difference(level_1='Control', level_2='Test')
diff_summary_2 = ttest.difference(level_1='Test', level_2='Control')
assert (diff_summary_1['p-value'].iloc[0] ==
diff_summary_2['p-value'].iloc[0])
assert (np.isclose(diff_summary_1['p-value'].iloc[0], .03334, 0.01))
def setup(self):
np.random.seed(123)
self.data = pd.DataFrame({
'variation_name':
['test', 'control', 'test2', 'test', 'control', 'test2'],
'success': [500, 42, 1005, 50, 4, 100],
'total': [1009, 104, 1502, 100, 10, 150],
'country': [
'us',
'us',
'us',
'gb',
'gb',
'gb',
]
})
self.test = spotify_confidence.StudentsTTest(
self.data,
numerator_column='success',
numerator_sum_squares_column='success',
denominator_column='total',
categorical_group_columns=['country', 'variation_name'])
def setup(self):
self.data = pd.DataFrame({
'variation_name': [
'test',
'control',
'test2',
'test',
'control',
'test2',
'test',
'control',
'test2',
'test',
'control',
'test2',
'test',
'control',
'test2',
'test',
'control',
'test2',
'test',
'control',
'test2',
'test',
'control',
'test2',
'test',
'control',
'test2',
'test',
'control',
'test2',
],
'nr_of_items': [
500,
8,
100,
510,
8,
100,
520,
9,
104,
530,
7,
100,
530,
8,
103,
500,
8,
100,
510,
8,
100,
520,
9,
104,
530,
7,
100,
530,
8,
103,
],
'nr_of_items_sumsq': [
1010,
32,
250,
1000,
30,
253,
1030,
33,
254,
1000,
30,
250,
1040,
31,
255,
1010,
22,
150,
1000,
20,
153,
1030,
23,
154,
1000,
20,
150,
1040,
21,
155,
],
'users': [
2010,
42,
250,
2000,
40,
253,
2030,
43,
254,
2000,
40,
250,
2040,
41,
255,
1010,
22,
150,
1000,
20,
153,
1030,
23,
154,
1000,
20,
150,
1040,
21,
155,
],
'days_since_reg': [
1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 1, 1, 1, 2, 2, 2,
3, 3, 3, 4, 4, 4, 5, 5, 5
],
'country': [
'us',
'us',
'us',
'us',
'us',
'us',
'us',
'us',
'us',
'us',
'us',
'us',
'us',
'us',
'us',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
'gb',
]
})
self.test = spotify_confidence.StudentsTTest(
self.data,
numerator_column='nr_of_items',
numerator_sum_squares_column='nr_of_items_sumsq',
denominator_column='users',
categorical_group_columns=['variation_name', 'country'],
ordinal_group_column='days_since_reg')