MLModelConfig

a model config defines execution properties within a pipeline.
usually to define how a pipeline should train and predict with your model.

Attributes

Name	Type	Description
by_twin	bool	define if pipeline need to train and use different model by twin item.
by_property	bool	define if pipeline need to train and use different model based on 'property_name'.
features	list	datapoint list to refine columns if necessary.
function	str	name of the function used instead of 'predict' on model inference.
model_key	str	key of the model to store (or use story property if dynamic).
model_type	str	reserved for 'mlflow' defines model flavor such as 'sklearn' or 'pyfunc'.
model_alias	str	reserved for 'mlflow' set the alias to target and use inside a pipeline.
properties_mapping	dict	dict to map properties expected by the script with the properties from the context.
property_name	str	define which property is looked for when using 'by_property'.
source	str	define name of model storage to use 'wizata'.
train_script	str	name of function referencing the script to train the model.
train_test_split_pct	float	percentage repartition to split the data for training and scoring.
train_test_split_type	str	type of splitting desired to split the train dataframe.
output_append	bool	true - default - append output to input dataframe. false to retrieve only 'predict' column(s).
output_columns_names	list	name list to rename columns inside output dataframe.
output_prefix	str	set a prefix to put before output column names (default or custom).

return current object id on Web API format.

return endpoint name used to contact backend.

load the object from a dict originating of a JSON format.

Name	Type	Default	Description
obj			object to load information from.

determine if configuration possess a target feature

specify the id_value neutrally

Name	Type	Default	Description
id_value

return:

transform current object into a dict that could be JSONIFY.

Name	Type	Default	Description
target	str	None

return: dumpable dict.