This repository was archived by the owner on Jun 30, 2022. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 10
/
Copy pathexperiment_client_test.py
170 lines (148 loc) · 7.75 KB
/
experiment_client_test.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
'''
ExperimentClient tests.
'''
import os
import unittest
import pandas as pd
import time
from mljar.client.project import ProjectClient
from mljar.client.dataset import DatasetClient
from mljar.client.experiment import ExperimentClient
from .project_based_test import ProjectBasedTest, get_postfix
class ExperimentClientTest(ProjectBasedTest):
def setUp(self):
proj_title = 'Test project-01'+get_postfix()
proj_task = 'bin_class'
self.expt_title = 'Test experiment-01'
self.validation_kfolds = 5
self.validation_shuffle = True
self.validation_stratify = True
self.validation_train_split = None
self.algorithms = ['xgb']
self.metric = 'logloss'
self.tuning_mode = 'Normal'
self.time_constraint = 1
self.create_enseble = False
# setup project
self.project_client = ProjectClient()
self.project = self.project_client.create_project(title = proj_title, task = proj_task)
# add training data
df = pd.read_csv('tests/data/test_1.csv')
cols = ['sepal length', 'sepal width', 'petal length', 'petal width']
target = 'class'
dc = DatasetClient(self.project.hid)
self.dataset = dc.add_dataset_if_not_exists(df[cols], df[target])
def tearDown(self):
# wait before clean, to have time to initialize models
time.sleep(60)
# clean
self.project_client.delete_project(self.project.hid)
def test_create_with_kfold_cv(self):
#Create experiment test with k-fold CV.
# add experiment
ec = ExperimentClient(self.project.hid)
self.assertNotEqual(ec, None)
# there should be none experiments
experiments = ec.get_experiments()
self.assertEqual(experiments, [])
# create new experiment
experiment = ec.add_experiment_if_not_exists(self.dataset, None, self.expt_title, self.project.task,
self.validation_kfolds, self.validation_shuffle,
self.validation_stratify, self.validation_train_split,
self.algorithms, self.metric,
self.tuning_mode, self.time_constraint, self.create_enseble)
self.assertNotEqual(experiment, None)
self.assertEqual(experiment.title, self.expt_title)
self.assertEqual(experiment.validation_scheme, "5-fold CV, Shuffle, Stratify")
self.assertEqual(experiment.metric, self.metric)
# get all experiments, should be only one
experiments = ec.get_experiments()
self.assertEqual(len(experiments), 1)
# get experiment by hid, there should be the same
experiment_2 = ec.get_experiment(experiment.hid)
self.assertEqual(experiment_2.hid, experiment.hid)
self.assertEqual(experiment_2.title, experiment.title)
self.assertEqual(experiment_2.metric, experiment.metric)
self.assertEqual(experiment_2.validation_scheme, experiment.validation_scheme)
self.assertTrue(experiment.equal(experiment_2))
# test __str__ method
self.assertTrue('id' in str(experiment_2))
self.assertTrue('title' in str(experiment_2))
self.assertTrue('metric' in str(experiment_2))
self.assertTrue('validation' in str(experiment_2))
def test_create_with_train_split(self):
#Create experiment with validation by train split.
# add experiment
ec = ExperimentClient(self.project.hid)
self.assertNotEqual(ec, None)
# there should be none experiments
experiments = ec.get_experiments()
self.assertEqual(experiments, [])
# create new experiment
experiment = ec.add_experiment_if_not_exists(self.dataset, None, self.expt_title, self.project.task,
self.validation_kfolds, self.validation_shuffle,
self.validation_stratify, 0.72,
self.algorithms, self.metric,
self.tuning_mode, self.time_constraint, self.create_enseble)
self.assertNotEqual(experiment, None)
self.assertEqual(experiment.title, self.expt_title)
self.assertEqual(experiment.validation_scheme, "Split 72/28, Shuffle, Stratify")
def test_create_with_validation_dataset(self):
#Create experiment with validation with dataset.
# add vald dataset
cols = ['sepal length', 'sepal width', 'petal length', 'petal width']
target = 'class'
df = pd.read_csv('tests/data/test_1_vald.csv')
dc = DatasetClient(self.project.hid)
vald_dataset = dc.add_dataset_if_not_exists(df[cols], df[target])
# add experiment
ec = ExperimentClient(self.project.hid)
self.assertNotEqual(ec, None)
# there should be none experiments
experiments = ec.get_experiments()
self.assertEqual(experiments, [])
# create new experiment
experiment = ec.add_experiment_if_not_exists(self.dataset, vald_dataset, self.expt_title, self.project.task,
self.validation_kfolds, self.validation_shuffle,
self.validation_stratify, 0.72,
self.algorithms, self.metric,
self.tuning_mode, self.time_constraint, self.create_enseble)
self.assertNotEqual(experiment, None)
self.assertEqual(experiment.title, self.expt_title)
self.assertEqual(experiment.validation_scheme, "With dataset")
def test_create_if_exists(self):
#Create experiment after experiment is already in project.
# add experiment
ec = ExperimentClient(self.project.hid)
self.assertNotEqual(ec, None)
# there should be none experiments
experiments = ec.get_experiments()
self.assertEqual(experiments, [])
# create new experiment
experiment = ec.add_experiment_if_not_exists(self.dataset, None, self.expt_title, self.project.task,
self.validation_kfolds, self.validation_shuffle,
self.validation_stratify, self.validation_train_split,
self.algorithms, self.metric,
self.tuning_mode, self.time_constraint, self.create_enseble)
self.assertNotEqual(experiment, None)
# get all experiments, should be only one
experiments = ec.get_experiments()
self.assertEqual(len(experiments), 1)
# try to create the same experiment
experiment_2 = ec.add_experiment_if_not_exists(self.dataset, None, self.expt_title, self.project.task,
self.validation_kfolds, self.validation_shuffle,
self.validation_stratify, self.validation_train_split,
self.algorithms, self.metric,
self.tuning_mode, self.time_constraint, self.create_enseble)
self.assertNotEqual(experiment, None)
# get all experiments, should be only one
experiments = ec.get_experiments()
self.assertEqual(len(experiments), 1)
# both should be the same
self.assertEqual(experiment_2.hid, experiment.hid)
self.assertEqual(experiment_2.title, experiment.title)
self.assertEqual(experiment_2.metric, experiment.metric)
self.assertEqual(experiment_2.validation_scheme, experiment.validation_scheme)
self.assertTrue(experiment.equal(experiment_2))
if __name__ == "__main__":
unittest.main()