Skip to content

Commit 96bb44c

Browse files
ludoviccludovicc
authored
Merge pull request LREN-CHUV#49 from LREN-CHUV/fix-no-valus
return 0/nan values when SQL results in zero rows
2 parents 42eb638 + 835ab0b commit 96bb44c

3 files changed

Lines changed: 71 additions & 173 deletions

File tree

‎python-summary-statistics/statistics.py‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -95,7 +95,7 @@ def intermediate_stats():
9595
types = _get_types(indep_vars + [dep_var])
9696

9797
if len(dep_var['series']) == 0:
98-
raise errors.UserError('Dependent variable has no values, check your SQL query.')
98+
logging.warning('Dependent variable has no values, check your SQL query.')
9999

100100
# Load data into a Pandas dataframe
101101
logging.info("Loading data...")

‎python-summary-statistics/tests/unit/fixtures.py‎

Lines changed: 0 additions & 124 deletions
This file was deleted.

‎python-summary-statistics/tests/unit/test_statistics.py‎

Lines changed: 70 additions & 48 deletions
Original file line numberDiff line numberDiff line change
@@ -1,78 +1,73 @@
11
import mock
22
import json
3-
from . import fixtures as fx
3+
from mip_helper import testing as t
4+
45
from statistics import intermediate_stats, aggregate_stats
56

67

78
@mock.patch('statistics.io_helper.fetch_data')
89
@mock.patch('statistics.io_helper.save_results')
910
def test_intermediate_stats_real(mock_save_results, mock_fetch_data):
1011
# input data with some null values
11-
data = fx.inputs_regression(include_categorical=True, add_null=True)
12-
data['data']['dependent'][0]['series'][0] = None
13-
data['data']['independent'][1]['series'][0] = None
12+
data = t.inputs_regression(include_nominal=True, add_null=True)
1413

1514
mock_fetch_data.return_value = data
1615

1716
intermediate_stats()
1817
results = json.loads(mock_save_results.call_args[0][0])
19-
assert len(results['data']) == 16
20-
assert results['data'][:2] == [
18+
assert len(results['data']) == 12
19+
res = sorted(results['data'], key=lambda d: (d['index'], tuple(d['group'])))
20+
assert t.round_dict(res[:2]) == [
2121
{
2222
'index': 'agegroup',
23-
'label': 'Age group',
23+
'label': 'Age Group',
2424
'type': 'polynominal',
2525
'group': ['-50y'],
26-
'group_variables': ['Age group'],
26+
'group_variables': ['Age Group'],
2727
'count': 3,
2828
'unique': 1,
2929
'top': '-50y',
3030
'frequency': {
3131
'-50y': 3,
32-
'59y-': 0,
3332
'50-59y': 0
3433
},
3534
'null_count': 0
3635
}, {
37-
'index': 'iq',
38-
'label': 'IQ',
39-
'type': 'real',
40-
'group': ['-50y'],
41-
'group_variables': ['Age group'],
42-
'count': 3,
43-
'mean': 73.8895774452,
44-
'std': 0.1412026822,
45-
'min': 73.7897320711,
46-
'25%': 73.8396547582,
47-
'50%': 73.8895774452,
48-
'75%': 73.9395001323,
49-
'max': 73.9894228193,
50-
'EX^2': 5459.6796241289,
51-
'null_count': 1
36+
'index': 'agegroup',
37+
'label': 'Age Group',
38+
'type': 'polynominal',
39+
'group': ['50-59y'],
40+
'group_variables': ['Age Group'],
41+
'count': 2,
42+
'unique': 1,
43+
'top': '50-59y',
44+
'frequency': {
45+
'50-59y': 2,
46+
'-50y': 0
47+
},
48+
'null_count': 0
5249
}
5350
]
5451

5552

5653
@mock.patch('statistics.io_helper.fetch_data')
5754
@mock.patch('statistics.io_helper.save_results')
5855
def test_intermediate_stats_nominal(mock_save_results, mock_fetch_data):
59-
# input data with some null values
60-
data = fx.inputs_classification(include_categorical=True)
61-
# data['data']['dependent'][0]['series'][0] = None
62-
# data['data']['independent'][1]['series'][0] = None
56+
data = t.inputs_classification(include_nominal=True)
6357

6458
mock_fetch_data.return_value = data
6559

6660
intermediate_stats()
6761
results = json.loads(mock_save_results.call_args[0][0])
68-
assert len(results['data']) == 16
69-
assert results['data'][:2] == [
62+
assert len(results['data']) == 12
63+
res = sorted(results['data'], key=lambda d: (d['index'], tuple(d['group'])))
64+
assert t.round_dict(res[:2]) == [
7065
{
7166
'index': 'adnicategory',
7267
'label': 'ADNI category',
7368
'type': 'polynominal',
7469
'group': ['-50y'],
75-
'group_variables': ['Age group'],
70+
'group_variables': ['Age Group'],
7671
'count': 3,
7772
'unique': 3,
7873
'top': 'Other',
@@ -83,18 +78,18 @@ def test_intermediate_stats_nominal(mock_save_results, mock_fetch_data):
8378
},
8479
'null_count': 0
8580
}, {
86-
'index': 'agegroup',
87-
'label': 'Age group',
81+
'index': 'adnicategory',
82+
'label': 'ADNI category',
8883
'type': 'polynominal',
89-
'group': ['-50y'],
90-
'group_variables': ['Age group'],
91-
'count': 3,
84+
'group': ['50-59y'],
85+
'group_variables': ['Age Group'],
86+
'count': 2,
9287
'unique': 1,
93-
'top': '-50y',
88+
'top': 'Other',
9489
'frequency': {
95-
'-50y': 3,
96-
'59y-': 0,
97-
'50-59y': 0
90+
'Other': 2,
91+
'CN': 0,
92+
'AD': 0
9893
},
9994
'null_count': 0
10095
}
@@ -103,17 +98,44 @@ def test_intermediate_stats_nominal(mock_save_results, mock_fetch_data):
10398

10499
@mock.patch('statistics.io_helper.fetch_data')
105100
@mock.patch('statistics.io_helper.save_results')
106-
@mock.patch('statistics.io_helper.save_error')
107-
def test_intermediate_stats_empty(mock_save_error, mock_save_results, mock_fetch_data):
108-
data = fx.inputs_regression(include_categorical=True, add_null=True)
109-
data['data']['dependent'][0]['series'] = []
101+
def test_intermediate_stats_empty(mock_save_results, mock_fetch_data):
102+
data = t.inputs_regression(include_nominal=True, limit_to=0)
110103
mock_fetch_data.return_value = data
111104

112-
with mock.patch('sys.exit'):
113-
intermediate_stats()
105+
intermediate_stats()
114106

115-
error = mock_save_error.call_args[0]
116-
assert error == ('Dependent variable has no values, check your SQL query.',)
107+
results = json.loads(mock_save_results.call_args[0][0])
108+
assert len(results['data']) == 12
109+
res = sorted(results['data'], key=lambda d: (d['index'], tuple(d['group'])))
110+
assert t.round_dict(res[:2]) == [
111+
{
112+
'index': 'agegroup',
113+
'label': 'Age Group',
114+
'type': 'polynominal',
115+
'group': ['-50y'],
116+
'group_variables': ['Age Group'],
117+
'count': 0,
118+
'unique': 0,
119+
'frequency': {
120+
'50-59y': 0,
121+
'-50y': 0
122+
},
123+
'null_count': 0
124+
}, {
125+
'index': 'agegroup',
126+
'label': 'Age Group',
127+
'type': 'polynominal',
128+
'group': ['50-59y'],
129+
'group_variables': ['Age Group'],
130+
'count': 0,
131+
'unique': 0,
132+
'frequency': {
133+
'50-59y': 0,
134+
'-50y': 0
135+
},
136+
'null_count': 0
137+
}
138+
]
117139

118140

119141
def intermediate_data_1():

0 commit comments

Comments
 (0)