def setUp(self):
name = "grid search test"
author = "srinidhi"
description = "Grid search cross validation - 3 folds"
syncer_obj = SyncerTest(
NewOrExistingProject(name, author, description),
DefaultExperiment(),
NewExperimentRun("Abc"),
ThriftConfig(None, None))
X = pd.DataFrame(np.random.randint(
0, 100, size=(2000, 4)), columns=list('ABCD'))
y = pd.DataFrame(np.random.randint(
0, 100, size=(2000, 1)), columns=['output'])
# Add tag for dataframe
syncer_obj.add_tag(X, "digits-dataset")
syncer_obj.clear_buffer()
tuned_parameters = [{'kernel': ['rbf'], 'gamma': [1e-3, 1e-4],
'C': [10, 100]}]
clf = GridSearchCV(SVC(), tuned_parameters, cv=3)
y = y.values.ravel()
clf.fit_sync(X, y)
events = syncer_obj.sync()
self.grid_search_event = events[0]
def setUp(self):
name = "logistic-test"
author = "srinidhi"
description = "income-level logistic regression"
syncer_obj = SyncerTest(
NewOrExistingProject(name, author, description),
DefaultExperiment(), NewExperimentRun("Abc"),
ThriftConfig(None, None))
# Creating the pipeline
pca = decomposition.PCA()
lr = linear_model.LinearRegression()
pipe = Pipeline(steps=[('pca', pca), ('logistic', lr)])
model = linear_model.LinearRegression()
np.random.seed(0)
X = pd.DataFrame(np.random.randint(0, 100, size=(100, 2)),
columns=list('AB'))
y = pd.DataFrame(np.random.randint(0, 100, size=(100, 1)),
columns=['output'])
# Add tags for models / dataframes
syncer_obj.add_tag(X, "digits-dataset")
syncer_obj.add_tag(pipe, "pipeline with pca + logistic")
syncer_obj.add_tag(pca, "decomposition PCA")
syncer_obj.add_tag(lr, "basic linear reg")
syncer_obj.clear_buffer()
pipe.fit_sync(X, y)
events = syncer_obj.sync()
self.pipeline_event = events[0]
def setUp(self):
name = "random split test"
author = "srinidhi"
description = "70/30 split"
syncer_obj = SyncerTest(
NewOrExistingProject(name, author, description),
DefaultExperiment(),
NewExperimentRun("Abc"),
ThriftConfig(None, None))
X = pd.DataFrame(np.random.randint(
0, 100, size=(100, 4)), columns=list('ABCD'))
y = pd.DataFrame(np.random.randint(
0, 100, size=(100, 1)), columns=['output'])
# Add tag for dataframe
syncer_obj.add_tag(X, "digits-dataset")
seed = 1
weights = [0.7, 0.3]
syncer_obj.clear_buffer()
x_train, x_test, y_train, y_test = (
cross_validation.train_test_split_sync(
X, y, train_size=0.7))
events = syncer_obj.sync()
self.random_split_event = events[0]
def setUpClass(self):
syncer_obj = SyncerTest(NewOrExistingProject("name", "author", "desc"),
NewOrExistingExperiment("expName", "expDesc"),
NewExperimentRun("expRunDesc"),
ThriftConfig(None, None))
events = syncer_obj.sync()
self.project_event = events[0]
self.experiment_event = events[1]
self.experiment_run_event = events[2]
def setUpClass(self):
name = "logistic-test"
author = "srinidhi"
description = "income-level logistic regression"
syncer_obj = SyncerTest(
NewOrExistingProject(name, author, description),
DefaultExperiment(),
NewExperimentRun("Abc"),
ThriftConfig(None, None),
)
events = syncer_obj.sync()
self.project_event = events[0]
self.experiment_event = events[1]
self.experiment_run_event = events[2]
def setUpClass(self):
name = "logistic-test"
author = "srinidhi"
description = "income-level logistic regression"
syncer_obj = SyncerTest(
NewOrExistingProject(name, author, description),
DefaultExperiment(),
NewExperimentRun("Abc"),
ThriftConfig(None, None))
letters = ['A', 'B', 'C', 'D']
X = np.random.choice(letters, size=(100, 1)).ravel()
model = preprocessing.LabelEncoder()
# Add tag for model
syncer_obj.add_tag(model, "label encoder")
syncer_obj.clear_buffer()
model.fit_transform_sync(X)
events = syncer_obj.sync()
self.fit_event = events[0]
self.transform_event = events[1]
def setUp(self):
name = "logistic-test"
author = "srinidhi"
description = "income-level logistic regression"
syncer_obj = SyncerTest(
NewOrExistingProject(name, author, description),
DefaultExperiment(), NewExperimentRun("Abc"),
ThriftConfig(None, None))
model = linear_model.LinearRegression()
np.random.seed(0)
X = pd.DataFrame(np.random.randint(0, 100, size=(100, 4)),
columns=list('ABCD'))
y = pd.DataFrame(np.random.randint(0, 100, size=(100, 1)),
columns=['output'])
# Add tags for models / dataframes
syncer_obj.add_tag(X, "digits-dataset")
syncer_obj.add_tag(model, "linear reg")
syncer_obj.clear_buffer()
model.fit_sync(X, y)
events = syncer_obj.sync()
self.fit_event = events[0]