问题描述
RandomizedSearchCV很有用，因为它不会尝试列出的所有参数。相反，它显示了一些并对其进行了测试以查看哪个更好。
但是我怎么知道测试了哪些参数？
例如，在下面的脚本中，测试了n_estimators，max_features和max_depth的哪些组合？ n_estimator = 10经过测试了吗？ n_estimator = 100经过测试了吗？
rf = RandomForestRegressor()
n_estimators = [int(x) for x in np.linspace(start=10,stop=2000,num=200)]
max_features = ["auto","sqrt","log2"]
max_depth = [int(x) for x in np.linspace(5,500,num=100)]
random_grid = {
"n_estimators": n_estimators,"max_features": max_features,"max_depth": max_depth,}
randomsearch = RandomizedSearchCV(rf,param_distributions=random_grid,cv=5)
randomsearch.fit(X_train,y_train)
解决方法
属性class DummyTestClass(unittest.TestCase):
my_guard = False
is_this_the_cicd_jenkins_machine = SomeClass.is_this_the_cicd_jenkins_machine()
@attr(testsuite='SUITE_1')
def test_a_change_my_guard(self):
# purpose of test is only to change the guard so that test b will be run if testsute SUITE_1 is selected form Jenkins
print('test a')
globals()['my_guard'] = True
# my_guard = True # also trying this,still my_guard is still false in test_b
@unittest.skipIf(my_guard is False,'skip test_b')
# @unittest.skipIf(globals()['my_guard'] is False,'skip test b') compilation error
# @unittest.skipIf(is_this_the_cicd_jenkins_machine,‘script skipped’) # ugly solution
@attr(testsuite='SUITE_1')
def test_b(self):
# controlled by guard
print('test b')
@attr(testsuite='SUITE_1')
def test_c_to_be_run_in_suite_1(self):
print('test c')
中提供了许多有关搜索的信息。将该dict导入数据框，您将为每个测试的超参数组合获得一行，其中包含超参数值，折叠和平均得分，可选的训练得分，训练时间等。