def test_get_config_override_profile():
config = DatabricksConfig.from_token("yo", "lo")
update_and_persist_config(TEST_PROFILE, config)
try:
provider = ProfileConfigProvider(TEST_PROFILE)
set_config_provider(provider)
config = get_config()
assert config.host == "yo"
assert config.token == "lo"
finally:
set_config_provider(None)
def test_get_config_override_custom():
class TestConfigProvider(DatabricksConfigProvider):
def get_config(self):
return DatabricksConfig.from_token("Override", "Token!")
try:
provider = TestConfigProvider()
set_config_provider(provider)
config = get_config()
assert config.host == "Override"
assert config.token == "Token!"
finally:
set_config_provider(None)
def test_get_config_bad_override():
with pytest.raises(Exception):
set_config_provider("NotAConfigProvider")
# MAGIC
# MAGIC Now train a binary classifier (using random forests) for predciting the target condition. We use MLFlow for tracking and registering the model, and use hyperopt for distributed hyper parameter tuning.
# COMMAND ----------
# DBTITLE 0,Temp fix for mlflow error message through load balancer
from databricks_cli.configure.provider import get_config_provider, set_config_provider, DatabricksConfigProvider
base_provider = get_config_provider()
class DynamicConfigProvider(DatabricksConfigProvider):
def get_config(self):
base_config = base_provider.get_config()
base_config.insecure = True
return base_config
set_config_provider(DynamicConfigProvider())
import urllib3
urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
# COMMAND ----------
from pyspark.ml.classification import LogisticRegression
# Create initial LogisticRegression model
lr = LogisticRegression(labelCol="label", featuresCol="features", maxIter=10)
# Train model with Training Data
lrModel = lr.fit(training_encounters)