def pi_init(self, str_xml: str):
"""
Handles configuration based on the GUI.
Called when the Alteryx engine is ready to provide the tool configuration from the GUI.
:param str_xml: The raw XML from the GUI.
"""
# Getting the dataName data property from the Gui.html
self.data_source = Et.fromstring(str_xml).find(
'DataImportSource').text if 'DataImportSource' in str_xml else None
self.data_location = Et.fromstring(str_xml).find(
'DataLocation').text if 'DataLocation' in str_xml else None
self.do_auto_viz = Et.fromstring(str_xml).find(
'DoAutoViz').text == "True" if 'DoAutoViz' in str_xml else False
self.display_browser = Et.fromstring(str_xml).find(
'DisplayBrowser'
).text == "True" if 'DisplayBrowser' in str_xml else False
self.use_https = Et.fromstring(str_xml).find(
'UseHttps').text == "True" if 'UseHttps' in str_xml else False
verify = False if self.use_https else True
# Get connection information
base = Et.fromstring(str_xml)
self.address = base.find(
'Address'
).text if 'Address' in str_xml else 'http://prerelease.h2o.ai'
username = base.find(
'Username').text if 'Username' in str_xml else 'h2oai'
password = self.alteryx_engine.decrypt_password(
base.find('Password').text,
0) if 'Password' in str_xml else 'h2oai'
self.dai = h2oai_client.Client(address=self.address,
username=username,
password=password,
verify=verify)
# Valid target name checks.
if self.data_source is None:
self.display_error_msg(
self.alteryx_engine.xmsg(
'Data Source must be specified. Please select a data source.'
))
elif self.data_location is None and self.data_source != "upload":
self.display_error_msg(
self.alteryx_engine.xmsg(
'Data Location must be specified. Please enter a data location'
))
error_msg = self.msg_str(self.str_file_path)
if error_msg != '':
self.alteryx_engine.output_message(self.n_tool_id,
Sdk.EngineMessageType.error,
self.xmsg(error_msg))
else:
self.is_valid = True
self.output_dataset = self.output_anchor_mgr.get_output_anchor(
'Dataset')
def process(dai_host, dai_user, dai_pass, train_ds, exp_config, project_name,
test_ds):
"""
:param dai_host: Driverless AI host URL e.g. http://localhost:12345
:param dai_user: Driverless AI user name
:param dai_pass: Driverless AI password
:param train_ds: path to training dataset csv file
:param exp_config: path to experiment config json file
:param project_name: Project name to organize datasets and experiments
:param test_ds: path to testing dataset csv file (optional)
:return: None
"""
# print all the passed parameters
# import inspect
# _, _, _, values = inspect.getargvalues(inspect.currentframe())
# print(values)
# Create a connection to Driverless AI
con = h2o.Client(address=dai_host, username=dai_user, password=dai_pass)
# Get project key
project_key = get_project_key(con, project_name)
# Upload datasets and link to project
test_ds_key = None
train_ds_key = upload_dataset_to_project(con, project_key, train_ds,
"Training")
if test_ds is not None:
test_ds_key = upload_dataset_to_project(con, project_key, test_ds,
"Testing")
# Read experiment config file and overwrite needed configs, save the config on file system
with open(exp_config, 'r') as read_file:
experiment_configs = json.load(read_file)
experiment_configs['dataset_key'] = train_ds_key
if test_ds_key is not None:
experiment_configs['testset_key'] = test_ds_key
with open('experiment-config.json', 'w') as write_file:
json.dump(experiment_configs, write_file, indent=4)
# Execute the experiment, link to project
experiment: h2o.Model = con.start_experiment_sync(**experiment_configs)
con.link_experiment_to_project(project_key, experiment.key)
# build mojo pipeline
mojo: h2o.MojoPipeline = con.build_mojo_pipeline_sync(experiment.key)
# download mojo and python scoring pipelines and experiment summary
con.download(experiment.scoring_pipeline_path, "")
con.download(experiment.summary_path, "")
con.download(mojo.file_path, "")
# Finally save experiment.json
with open('experiment.json', 'w') as write_file:
json.dump(experiment.dump(), write_file, indent=4)
def pi_init(self, str_xml: str):
"""
Handles configuration based on the GUI.
Called when the Alteryx engine is ready to provide the tool configuration from the GUI.
:param str_xml: The raw XML from the GUI.
"""
# Getting the dataName data property from the Gui.html
self.model_id = Et.fromstring(str_xml).find(
'ModelID').text if 'ModelID' in str_xml else None
self.is_timeseries = Et.fromstring(str_xml).find(
'IsTimeSeries'
).text == "True" if "IsTimeSeries" in str_xml else False
self.target_col = Et.fromstring(str_xml).find(
"TargetColumn").text if "TargetColumn" in str_xml else ""
self.dataset_key = Et.fromstring(str_xml).find(
'DatasetKey').text if "DatasetKey" in str_xml else ""
self.display_browser = Et.fromstring(str_xml).find(
'DisplayBrowser'
).text == "True" if "DisplayBrowser" in str_xml else False
self.use_https = Et.fromstring(str_xml).find(
'UseHttps').text == "True" if 'UseHttps' in str_xml else False
verify = False if self.use_https else True
# Get connection information
base = Et.fromstring(str_xml)
self.addr = base.find(
'Address'
).text if 'Address' in str_xml else 'http://prerelease.h2o.ai'
username = base.find(
'Username').text if 'Username' in str_xml else 'h2oai'
password = self.alteryx_engine.decrypt_password(
base.find('Password').text,
0) if 'Password' in str_xml else 'h2oai'
self.dai = h2oai_client.Client(address=self.addr,
username=username,
password=password,
verify=verify)
# Valid target name checks.
error_msg = self.msg_str(self.str_file_path)
if error_msg != '':
self.alteryx_engine.output_message(self.n_tool_id,
Sdk.EngineMessageType.error,
self.xmsg(error_msg))
else:
self.is_valid = True
self.output_anchor = self.output_anchor_mgr.get_output_anchor('MLI')
def pi_init(self, str_xml: str):
"""
Handles configuration based on the GUI.
Called when the Alteryx engine is ready to provide the tool configuration from the GUI.
:param str_xml: The raw XML from the GUI.
"""
# Getting the dataName data property from the Gui.html
self.target_name = Et.fromstring(str_xml).find(
'TargetColumn').text if 'TargetColumn' in str_xml else None
#self.output_field = Et.fromstring(str_xml).find('OutputField').text if 'OutputField' in str_xml else None
self.accuracy = int(Et.fromstring(str_xml).find(
'Accuracy').text) if 'Accuracy' in str_xml else None
self.time = int(Et.fromstring(str_xml).find(
'Time').text) if 'Time' in str_xml else None
self.interpretability = int(
Et.fromstring(str_xml).find('Interpretability').text
) if 'Interpretability' in str_xml else None
self.is_classification = Et.fromstring(str_xml).find(
'IsClassification'
).text == 'True' if 'IsClassification' in str_xml else None
# Default function for DAI is to have Time Series off
self.is_timeseries = Et.fromstring(str_xml).find(
'IsTimeSeries'
).text == 'True' if 'IsTimeSeries' in str_xml else False
self.config_string = Et.fromstring(
str_xml).find('AdditionalUserConfigs'
).text if 'AdditionalUserConfigs' in str_xml else ""
reg_scorer = Et.fromstring(str_xml).find(
'RegressionScoringMetric'
).text if 'RegressionScoringMetric' in str_xml else ""
class_scorer = Et.fromstring(str_xml).find(
'ClassificationScoringMetric'
).text if 'ClassificationScoringMetric' in str_xml else ""
self.use_https = Et.fromstring(str_xml).find(
'UseHttps').text == "True" if 'UseHttps' in str_xml else False
verify = False if self.use_https else True
if self.is_classification:
self.scorer = class_scorer
else:
self.scorer = reg_scorer
self.display_browser = Et.fromstring(str_xml).find(
'DisplayBrowser'
).text == 'True' if 'DisplayBrowser' in str_xml else False
# Get connection information
base = Et.fromstring(str_xml)
self.addr = base.find(
'Address'
).text if 'Address' in str_xml else 'http://prerelease.h2o.ai'
username = base.find(
'Username').text if 'Username' in str_xml else 'h2oai'
password = self.alteryx_engine.decrypt_password(
base.find('Password').text,
0) if 'Password' in str_xml else 'h2oai'
self.dai = h2oai_client.Client(address=self.addr,
username=username,
password=password,
verify=verify)
# Valid target name checks.
if self.target_name is None:
self.display_error_msg(
self.alteryx_engine.xmsg(
'Target Column cannot be empty. Please enter a Target Column.'
))
elif len(self.target_name) > 255:
self.display_error_msg(
self.alteryx_engine.xmsg(
'Target Column cannot be greater than 255 characters.'))
error_msg = self.msg_str(self.str_file_path)
if error_msg != '':
self.alteryx_engine.output_message(self.n_tool_id,
Sdk.EngineMessageType.error,
self.xmsg(error_msg))
else:
self.is_valid = True
self.output_anchor = self.output_anchor_mgr.get_output_anchor(
'Predictions')
self.output_model = self.output_anchor_mgr.get_output_anchor('Model')