|
1
|
|
|
import pytest |
|
2
|
|
|
|
|
3
|
|
|
|
|
4
|
|
|
@pytest.fixture |
|
5
|
|
|
def data_manager_command_decorators(somagic): |
|
6
|
|
|
return { |
|
7
|
|
|
'data_manager_cmd': somagic.commands_decorators.data_manager_command, |
|
8
|
|
|
'arbitrary_cmd': somagic.commands_decorators.arbitrary_command, |
|
9
|
|
|
} |
|
10
|
|
|
|
|
11
|
|
|
|
|
12
|
|
|
@pytest.fixture |
|
13
|
|
|
def define_command(): |
|
14
|
|
|
def _define_engine_command(decorator, command_function): |
|
15
|
|
|
decorator(command_function) |
|
16
|
|
|
return _define_engine_command |
|
17
|
|
|
|
|
18
|
|
|
|
|
19
|
|
|
@pytest.fixture |
|
20
|
|
|
def get_command(somagic): |
|
21
|
|
|
def _get_command(command_name: str): |
|
22
|
|
|
return getattr(somagic.command, command_name) |
|
23
|
|
|
return _get_command |
|
24
|
|
|
|
|
25
|
|
|
|
|
26
|
|
|
@pytest.fixture |
|
27
|
|
|
def test_discretizer(): |
|
28
|
|
|
from so_magic.data.discretization import Discretizer, BinningAlgorithm |
|
29
|
|
|
|
|
30
|
|
|
alg = BinningAlgorithm.from_built_in('pd.cut') |
|
31
|
|
|
|
|
32
|
|
|
discretizer = Discretizer.from_algorithm(alg) |
|
33
|
|
|
return discretizer |
|
34
|
|
|
|
|
35
|
|
|
|
|
36
|
|
|
@pytest.fixture |
|
37
|
|
|
def discretize_command(): |
|
38
|
|
|
import pandas as pd |
|
39
|
|
|
|
|
40
|
|
|
def get_discretize_command(discretizer): |
|
41
|
|
|
def test_discretize_command(data_manager, datapoints, attribute, nb_bins, new_column_name): |
|
42
|
|
|
output = discretizer.discretize(datapoints, attribute, nb_bins) |
|
43
|
|
|
data_manager.datapoints.add_column(output['result'], new_column_name) |
|
44
|
|
|
return test_discretize_command |
|
45
|
|
|
return get_discretize_command |
|
46
|
|
|
|
|
47
|
|
|
|
|
48
|
|
|
@pytest.fixture |
|
49
|
|
|
def validate_discretization_operation_behaviour(): |
|
50
|
|
|
def _validate_discretization_operation(cmd, algorithm): |
|
51
|
|
|
datapoints = cmd.args[0] |
|
52
|
|
|
target_column = cmd.args[1] |
|
53
|
|
|
nb_bins = cmd.args[2] |
|
54
|
|
|
min_value = min(iter(datapoints.column(target_column))) |
|
55
|
|
|
max_value = max(iter(datapoints.column(target_column))) |
|
56
|
|
|
bin_size = (max_value - min_value) / float(nb_bins) |
|
57
|
|
|
computed_bins = algorithm.output['settings']['used_bins'] |
|
58
|
|
|
assert [_ for _ in computed_bins] == [-0.1, 25.0, 50.0, 75.0, 100.0] |
|
59
|
|
|
|
|
60
|
|
|
input_arguments = algorithm.output['settings']['arguments'] |
|
61
|
|
|
to_check = [len(input_arguments[0]), input_arguments[1]] |
|
62
|
|
|
assert to_check == [len(datapoints), nb_bins] |
|
63
|
|
|
assert type(datapoints.column(target_column)) == type(input_arguments[0]) |
|
64
|
|
|
assert list(datapoints.column(target_column)) == list(input_arguments[0]) |
|
65
|
|
|
# assert algorithm.output['settings']['parameters'] == [] |
|
66
|
|
|
return _validate_discretization_operation |
|
67
|
|
|
|
|
68
|
|
|
|
|
69
|
|
|
@pytest.fixture |
|
70
|
|
|
def discretiztion_test_data(somagic, load_test_data_this): |
|
71
|
|
|
load_test_data_this(somagic) |
|
72
|
|
|
print('DATAPOINTS BEFORE', len(somagic.datapoints.attributes)) |
|
73
|
|
|
print(set(type(x) for x in somagic.dataset.datapoints.column('Creative'))) |
|
74
|
|
|
series = somagic.dataset.datapoints.column('Creative').replace('', 0.0, inplace=False) |
|
75
|
|
|
assert all(type(x) == float for x in series) |
|
76
|
|
|
print(type(series)) |
|
77
|
|
|
print('MIN', min(series)) |
|
78
|
|
|
print('MAX', max(series)) |
|
79
|
|
|
|
|
80
|
|
|
somagic.datapoints.add_column(list(series), 'Creative') |
|
81
|
|
|
print('DATAPOINTS AFTER', len(somagic.datapoints.attributes)) |
|
82
|
|
|
print(set(type(x) for x in somagic.dataset.datapoints.column('Creative'))) |
|
83
|
|
|
|
|
84
|
|
|
assert all(type(x) == float for x in somagic.datapoints.observations['Creative']) |
|
85
|
|
|
|
|
86
|
|
|
return { |
|
87
|
|
|
'success': [ |
|
88
|
|
|
'Creative' |
|
89
|
|
|
], |
|
90
|
|
|
'fail': [ |
|
91
|
|
|
'Energetic' |
|
92
|
|
|
], |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
|
|
96
|
|
|
def test_discretization_operation(somagic, data_manager_command_decorators, discretiztion_test_data, define_command, get_command, test_discretizer, discretize_command, validate_discretization_operation_behaviour): |
|
97
|
|
|
print('INFO: datapoints columns:', somagic.datapoints.attributes) |
|
98
|
|
|
define_command(somagic.commands_decorators.data_manager_command(), discretize_command(test_discretizer)) |
|
99
|
|
|
print('ELA', set(type(x) for x in somagic.dataset.datapoints.column('Creative'))) |
|
100
|
|
|
for attr_name in discretiztion_test_data['success']: |
|
101
|
|
|
cmd = get_command('test_discretize_command') |
|
102
|
|
|
cmd.args = [somagic.datapoints, attr_name, 4, f'binned_{attr_name}'] |
|
103
|
|
|
cmd.execute() |
|
104
|
|
|
|
|
105
|
|
|
validate_discretization_operation_behaviour(cmd, test_discretizer.algorithm) |
|
106
|
|
|
|
|
107
|
|
|
for attr_name in discretiztion_test_data['fail']: |
|
108
|
|
|
cmd = get_command('test_discretize_command') |
|
109
|
|
|
cmd.args = [somagic.datapoints, attr_name, 4, f'binned_{attr_name}'] |
|
110
|
|
|
with pytest.raises(TypeError): |
|
111
|
|
|
cmd.execute() |
|
112
|
|
|
|
boromir674 /
so-magic
