das.helpers.schemas package

Submodules

das.helpers.schemas.generic module

class das.helpers.schemas.generic.ArrayFactory(array_list, title)

Bases: object

A helper class to create array-type schema using Series objects.

Helps IoT-TICKET to map input data into correct format.

Parameters
array_listlist

List of Series-type objects.

titlestr

Name for IoT-TICKET input node.

Methods

as_dict
array_list
as_dict()
title
class das.helpers.schemas.generic.ObjectFactory(object_list)

Bases: object

A helper class to be create object-type schema using Parameters and SeriesWithKey objects.

Helps IoT-TICKET to map input data into correct format.

Parameters
object_listlist

List of objects of type Parameter or SeriesWithKey.

Methods

as_dict
as_dict()
object_list
class das.helpers.schemas.generic.Parameter(key, title, item_type)

Bases: object

A helper class to be used together with ObjectFactory class.

Helps IoT-TICKET to map input parameter into correct format.

Parameters
keystr

Key under which the parameter value is mapped to.

titlestr

Name for IoT-TICKET input node.

item_typestr

Type of the input data. Must be ‘number’ or ‘string’.

Methods

as_dict
as_dict()
item_type
key
title
class das.helpers.schemas.generic.Series(length, item_type, title='Series')

Bases: object

A helper class to be used together with ArrayFactory class.

Helps IoT-TICKET to map input series into correct format.

Parameters
lengthint

Length of the series; i.e. how many items will be send to service. Maximum is 10 000 set by IoT-TICKET.

item_typestr

Type of the input data. Must be ‘number’ or ‘string’.

titlestr

Name for IoT-TICKET input node.

Methods

as_dict
as_dict()
item_type
length
title = 'Series'
class das.helpers.schemas.generic.SeriesWithKey(length, item_type, key='series', title='Series')

Bases: object

A helper class to be used together with ObjectFactory class.

Helps IoT-TICKET to map input series into correct format.

Parameters
lengthint

Length of the series; i.e. how many items will be send to service. Maximum is 10 000 set by IoT-TICKET.

item_type: str

Type of the input data. Must be ‘number’ or ‘string’.

keystr

Key under which the series is mapped to.

titlestr

Name for IoT-TICKET input node.

Methods

as_dict
as_dict()
item_type
key = 'series'
length
title = 'Series'

das.helpers.schemas.notebook module

class das.helpers.schemas.notebook.NotebookNode(name, type, binding_key)

Bases: object

A helper class to be used with notebook jobs’ inputs and outputs.

Parameters
namestr

Name of the node.

typestr

The type of the node. Must be one of ‘attribute_id’, ‘bool’, ‘number’, ‘string’.

binding_keystr

The key to bind the node to an actual input or output in IoT-TICKET. For example the attribute_id value.

Examples

>>> job = DatabricksJob.from_json(job_json, credential)
>>> job.notebook_inputs =  [NotebookNode('input series', 'attribute_id', 'some id here')]
>>> job.notebook_outputs = [NotebookNode('output series', 'attribute_id', 'some id2 here']

Methods

as_dict
as_dict()
binding_key
name
type

das.helpers.schemas.tensor module

class das.helpers.schemas.tensor.ColSpec(key, title, item_type, length=1)

Bases: object

A helper class to be used together with TensorInstances class.

Helps to map IoT-TICKET input data under the correct column.

Parameters
keystr

Name of the column to which the data is mapped.

titlestr

Name for IoT-TICKET input node.

item_typestr

Type of the input data. Must be one of ‘number’, ‘string’, ‘number_array’, ‘string_array’

lengthint

Length of the item. Should be at least 2 for number and string arrays.

Methods

as_dict
as_dict()
item_type
key
length = 1
title
class das.helpers.schemas.tensor.ServingOutput(key='predictions', title='Predictions', item_type='number_array')

Bases: object

ServingOutput class. Defines output schema for default response format for model serving’s predict API.

Default response format is:

{
    "predictions": <value>|<(nested)list>|<list-of-objects>
}

See more on https://www.tensorflow.org/tfx/serving/api_rest#response_format_4

Parameters
keystr

Key whose value contains the predictions. Defaults to ‘predictions’.

titlestr

Title for IoT-TICKET ml-service output for predictions.

item_typstr

Type of the item. In the sense of default response format, the corresponding possible values are ‘number’ or ‘string’ (<value>), ‘number_array’ or ‘string_array’ (<(nested)list>) and ‘object_list’ (<list-of-objects).

Methods

as_dict
as_dict()
item_type = 'number_array'
key = 'predictions'
title = 'Predictions'
class das.helpers.schemas.tensor.TensorInstances(columns, length=1)

Bases: object

TensorInstances class that helps to create a schema for IoT-TICKET so it can run inference against model serving using TF serving’s “instances” format.

https://mlflow.org/docs/latest/models.html#deploy-mlflow-models https://www.tensorflow.org/tfx/serving/api_rest#specifying_input_tensors_in_row_format

Parameters
columnslist

List of ColSpec instances.

lengthint

Number of instances (rows) to be used when running inference.

Methods

as_dict
as_dict()
columns
length = 1

Module contents