Client¶

Note

The client supports password and token based authentication, such as local, PAM, LDAP, and OpenID. Certificate based authentication such as Mutual TLS is not currently supported by the client.

API Reference:

class Client¶

Connect to and interact with a Driverless AI server.

Parameters:

address (str) – full URL of the Driverless AI server to connect to
username (Optional[str]) – username for authentication on the Driverless AI server
password (Optional[str]) – password for authentication on the Driverless AI server
token_provider (Optional[Callable[[], str]]) – callable that provides an authentication token, if provided, will ignore username and password values
verify (Union[bool, str]) – when using https on the Driverless AI server, setting this to False will disable SSL certificates verification. A path to cert(s) can also be passed to verify, see: https://requests.readthedocs.io/en/master/user/advanced/#ssl-cert-verification
backend_version_override (Optional[str]) – version of client backend to use, overrides Driverless AI server version detection. Specify "latest" to get the most recent backend supported. In most cases the user should rely on Driverless AI server version detection and leave this as the default None.

Examples:

### Connect with username and password
dai = driverlessai.Client(
    address='http://localhost:12345',
    username='py',
    password='py'
)

### Connect with token (assumes the Driverless AI server is configured
### to allow clients to authenticate through tokens)

# 1) set up a token provider with a refresh token from the Driverless AI web UI
token_provider = driverlessai.token_providers.OAuth2TokenProvider(
    refresh_token="eyJhbGciOiJIUzI1N...",
    client_id="python_client",
    token_endpoint_url="https://keycloak-server/auth/realms/..."
    token_introspection_url="https://keycloak-server/auth/realms/..."
)

# 2) use the token provider to get authorization to connect to the
# Driverless AI server
dai = driverlessai.Client(
    address="https://localhost:12345",
    token_provider=token_provider.ensure_fresh_token
)

property admin: driverlessai._admin.Admin¶

Performs administrative tasks on Driverless AI server.

Warning

Requires DriverlessAI server version 1.10.3 or higher.

Return type:	`Admin`

property autodocs: driverlessai._autodoc.AutoDocs¶

Interact with autodocs on the Driverless AI server.

Warning

A beta API that is subject to future changes.

Return type:	`AutoDocs`

property autoviz: driverlessai._autoviz.AutoViz¶

Interact with dataset visualizations on the Driverless AI server.

Return type:	`AutoViz`

property connectors: driverlessai._datasets.Connectors¶

Interact with connectors on the Driverless AI server.

Return type:	`Connectors`

property datasets: driverlessai._datasets.Datasets¶

Interact with datasets on the Driverless AI server.

Return type:	`Datasets`

property deployments: driverlessai._deployments.Deployments¶

Interact with datasets on the Driverless AI server.

Warning

Requires DriverlessAI server version 1.10.6 or higher.

Warning

A beta API that is subject to future changes.

Return type:	`Deployments`

property experiments: driverlessai._experiments.Experiments¶

Interact with experiments on the Driverless AI server.

Return type:	`Experiments`

property mli: driverlessai._mli.MLI¶

Interact with experiment interpretations on the Driverless AI server.

Return type:	`MLI`

property model_diagnostics: driverlessai._model_diagnostics.ModelDiagnostics¶

Interact with model diagnostics on the Driverless AI server.

Return type:	`ModelDiagnostics`

property projects: driverlessai._projects.Projects¶

Interact with projects on the Driverless AI server.

Return type:	`Projects`

property recipes: driverlessai._recipes.Recipes¶

Interact with recipes on the Driverless AI server.

Return type:	`Recipes`

property server: driverlessai._server.Server¶

Get information about the Driverless AI server.

Return type:	`Server`

Interact with Server Entities¶

The following attributes of a connected client allow you to interact with entities such as datasets or experiments on the Driverless AI server.

AutoViz¶

The autoviz attribute of a connected client provides methods to create, get, and list dataset visualizations on the Driverless AI server.

API Reference:

class AutoViz¶

Interact with dataset visualizations on the Driverless AI server.

create(dataset: Dataset) → Visualization¶

Create a dataset visualization on the Driverless AI server.

Parameters:	dataset (`Dataset`) – Dataset object
Return type:	`Visualization`

create_async(dataset: Dataset) → VisualizationJob¶

Launch creation of a dataset visualization on the Driverless AI server.

Parameters:	dataset (`Dataset`) – Dataset object
Return type:	`VisualizationJob`

get(key: str) → Visualization¶

Get a Visualization object corresponding to a dataset visualization on the Driverless AI server.

Parameters:	key (`str`) – Driverless AI server’s unique ID for the visualization
Return type:	`Visualization`

gui() → Hyperlink¶

Get full URL for the AutoViz page on the Driverless AI server.

Return type:	`Hyperlink`

list(start_index: int = 0, count: Optional[int] = None) → Sequence[Visualization]¶

Return list of dataset Visualization objects.

Parameters:	start_index (`int`) – index on Driverless AI server of first visualization in list count (`Optional`[`int`]) – number of visualizations to request from the Driverless AI server
Return type:	`Sequence`[`Visualization`]

Admin¶

The admin attribute of a connected client provides methods to perform administrative tasks on the Driverless AI server.

API Reference:

class Admin¶

property is_admin: Optional[bool]¶

Returns True if the user is an admin.

Return type:	`Optional`[`bool`]

list_current_users() → List[str]¶

Returns a list of users who are currently: logged-in to the Driverless AI server.

Warning

Requires DriverlessAI server version 1.10.5 or higher.

Warning

A beta API that is subject to future changes.

Return type:	`List`[`str`]

list_datasets(username: str) → List[DatasetProxy]¶

List datasets of the specified user.

Warning

A beta API that is subject to future changes.

Return type:	`List`[`DatasetProxy`]

list_experiments(username: str) → List[ExperimentProxy]¶

List experiments of the specified user.

Warning

A beta API that is subject to future changes.

Return type:	`List`[`ExperimentProxy`]

list_server_logs() → List[DAIServerLog]¶

Lists the server logs of Driverless AI server.

Warning

Requires DriverlessAI server version 1.10.5 or higher.

Return type:	`List`[`DAIServerLog`]

list_users() → List[str]¶

Returns a list of users in the Driverless AI server.

Return type:	`List`[`str`]

transfer_data(from_user: str, to_user: str) → None¶

Transfer all data of from_user to to_user.

Return type:	`None`

class DatasetProxy¶

A Proxy for admin access for a dataset in the Driverless AI server.

delete() → None¶

Delete this entity.

Return type:	`None`

property key: str¶

Universally unique identifier.

Return type:	`str`

property name: str¶

Display name.

Return type:	`str`

property owner: str¶

Owner of the object.

Return type:	`str`

class ExperimentProxy¶

A Proxy for admin access for an experiment in the Driverless AI server.

property creation_timestamp: float¶

Creation timestamp in seconds since the epoch (POSIX timestamp).

Return type:	`float`

property datasets: Dict[str, Optional[driverlessai._admin.DatasetProxy]]¶

Dictionary of train_dataset,``validation_dataset``, and test_dataset used for this experiment.

Return type:	`Dict`[`str`, `Optional`[`DatasetProxy`]]

delete() → None¶

Delete this entity.

Return type:	`None`

is_complete() → bool¶

Returns True if this job completed successfully.

Return type:	`bool`

is_running() → bool¶

Returns True if this job is scheduled, running, or finishing.

Return type:	`bool`

property key: str¶

Universally unique identifier.

Return type:	`str`

property name: str¶

Display name.

Return type:	`str`

property owner: str¶

Owner of the object.

Return type:	`str`

property run_duration: Optional[float]¶

Run duration in seconds.

Return type:	`Optional`[`float`]

property settings: Dict[str, Any]¶

Experiment settings.

Return type:	`Dict`[`str`, `Any`]

property size: int¶

Size in bytes of all experiment’s files on the Driverless AI server.

Return type:	`int`

status(verbose: int = 0) → str¶

Returns the status of this job.

Parameters:	verbose (`int`) – 0: short description 1: short description with progress percentage 2: detailed description with progress percentage
Return type:	`str`

Connectors¶

The connectors attribute of a connected client currently gives you a method to list connectors that have been enabled on the Driverless AI server.

Connectors are strings that designate a method to get data into the server:

upload : standard upload feature
file : local file system/server file system
hdfs : Hadoop file system, remember to configure the HDFS config folder path and keytab on the server
dtap : Blue Data Tap file system, remember to configure DTap on the server
s3 : Amazon S3, optionally configure secret and access key on the server
gcs : Google Cloud Storage, remember to configure gcs_path_to_service_account_json on the server
gbq : Google Big Query, remember to configure gcs_path_to_service_account_json on the server
minio : Minio Cloud Storage, remember to configure secret and access key on the server
snow : Snowflake Data Warehouse, remember to configure Snowflake credentials on the server (account name, username, password)
kdb : KDB+ Time Series Database, remember to configure KDB credentials on the server (hostname and port, optionally: username, password, classpath, and jvm_args)
azrbs : Azure Blob Storage, remember to configure Azure credentials on the server (account name, account key)
jdbc: JDBC Connector
hive: Hive Connector
recipe_file: custom recipe file upload
recipe_url: custom recipe upload via url
feature_store: Feature Store, remember to configure feature_store_endpoint_url on the server

These strings can be passed as the data_source argument for the dataset create functions. Data sources are exposed in the form of the file systems, and each file system is prefixed by a unique prefix. For example:

To reference data on S3, use s3://.
To reference data on HDFS, use the prefix hdfs://.
To reference data on Azure Blob Store, use https://<storage_name>.blob.core.windows.net.
To reference data on BlueData Datatap, use dtap://.
To reference data on Google BigQuery, make sure you know the Google BigQuery dataset and the table that you want to query. Use a standard SQL query to ingest data.
To reference data on Google Cloud Storage, use gs://.
To reference data on kdb+, use the hostname and the port http://<kdb_server>:<port>.
To reference data on Minio, use http://<endpoint_url>.
To reference data on Snowflake, use a standard SQL query to ingest data.
To access a SQL database via JDBC, use a SQL query with the syntax associated with your database.

Advanced Connectors¶

Advanced connectors require a data_source_config argument for the dataset create functions. The following connectors are considered advanced:

Google Big Query:

config = {
    'gbq_dataset_name': <required>,
    'gbq_bucket_name': <required>,
    'gbq_location': <optional>  # Specify a location/region to use with the GBQ connector.
    'gbq_project': <optional>  # Specify a project to use with the GBQ connector.
}

Note

Driverless AI’s connection to Google Big Query will inherit the top-level directory from the service JSON file. So if a dataset named “my-dataset” is in a top-level directory named “dai-gbq”, then the value for the dataset ID input field would be “my-dataset” and not “dai-gbq:my-dataset”.

Note

gbq_location is only supported in Driverless AI server versions >= 1.10.1.

Note

gbq_project is only supported in Driverless AI server versions >= 1.9.3.

Hive:

config = {
    'hive_conf_path': <optional>,  # Path to hadoop configuration directory.
    'hive_auth_type': <optional>,  # Type of authentication to use, can be [noauth, keytab, keytabimpersonation].
    'hive_keytab_path': <optional>,  # Path to keytab if using keytab authentication.
    'hive_principal_user': <optional>,  # User ID authorized by keytab to make queries.
    'hive_default_config': <optional>  # Name of database configuration in config.toml to use. If provided, will ignore all other optional arguments, and will take them directly from config.toml.
}

JDBC:

config = {
    'jdbc_username': <required>,
    'jdbc_password': <required>,
    'jdbc_url': <optional>,  # JDBC connection url. Requires jdbc_jar and jdbc_driver to be provided.
    'jdbc_jar': <optional>,  # Path to JDBC driver jar. Requires jdbc_url and jdbc_driver to be provided.
    'jdbc_driver': <optional>,  # Classpath of JDBC driver. Requires jdbc_jar and jdbc_url to be provided.
    'jdbc_default_config': <optional>,  # Name of database configuration in config.toml to use. If provided, will ignore jdbc_jar, jdbc_url, jdbc_driver values.
    'id_column': <optional>  # Name of id column in dataset. Specify this field when making large data queries.
}

Note

Due to resource sharing within Driverless AI, the JDBC Connector is only allocated a relatively small amount of memory. When making large queries, the ID column is used to partition the data into manageable portions. This ensures that the maximum memory allocation is not exceeded. If a query that is larger than the maximum memory allocation is made without specifying an ID column, the query will not complete successfully.

Snowflake:

config = {
    'snowflake_database': <required>,
    'snowflake_warehouse': <required>,
    'snowflake_schema': <required>,
    'snowflake_region': <optional>,
    'snowflake_role': <optional>,
    'snowflake_formatting': <optional>,  # Additional arguments for formatting the output SQL query to csv file. See snowflake documentation for "Create File Format".
    'snowflake_username': <optional>,
    'snowflake_password': <optional>
    'snowflake_account': <optional>
}

Note

snowflake_account is only supported in Driverless AI server versions >= 1.10.2.

Feature Store:

config = {
    'feature_store_project': <required>
}

Note

Feature Store connector is only supported in Driverless AI server versions >= 1.10.4.

API Reference:

class Connectors¶

Interact with data sources that are enabled on the Driverless AI server.

list() → List[str]¶

Return list of data sources enabled on the Driverless AI server.

Examples:

dai.connectors.list()

Return type:	`List`[`str`]

Datasets¶

The datasets attribute of a connected client provides methods to create, get, and list datasets on the Driverless AI server.

API Reference:

class Datasets¶

Interact with datasets on the Driverless AI server.

create(data: Union[str, pandas.DataFrame], data_source: str = 'upload', data_source_config: Dict[str, str] = None, force: bool = False, name: str = None) → Dataset¶

Create a dataset on the Driverless AI server and return a Dataset object corresponding to the created dataset.

Parameters:

data (Union[str, ForwardRef]) – path to data file(s) or folder, a Pandas DataFrame, or query string for SQL based data sources
data_source (str) – name of connector to use for data transfer (use driverlessai.connectors.list() to see configured names)
data_source_config (Dict[str, str]) – dictionary of configuration options for advanced connectors
force (bool) – create new dataset even if dataset with same name already exists
name (str) – dataset name on the Driverless AI server

Examples:

ds = dai.datasets.create(
    data='s3://h2o-public-test-data/smalldata/iris/iris.csv',
    data_source='s3'
)

Return type:	Dataset

create_async(data: Union[str, pandas.DataFrame], data_source: str = 'upload', data_source_config: Dict[str, str] = None, force: bool = False, name: str = None) → DatasetJob¶

Launch creation of a dataset on the Driverless AI server and return a DatasetJob object to track the status.

Parameters:

data (Union[str, ForwardRef]) – path to data file(s) or folder, a Pandas DataFrame, or query string for SQL based data sources
data_source (str) – name of connector to use for data transfer (use driverlessai.connectors.list() to see configured names)
data_source_config (Dict[str, str]) – dictionary of configuration options for advanced connectors
force (bool) – create new dataset even if dataset with same name already exists
name (str) – dataset name on the Driverless AI server

Examples:

ds = dai.datasets.create_async(
    data='s3://h2o-public-test-data/smalldata/iris/iris.csv',
    data_source='s3'
)

Return type:	DatasetJob

get(key: str) → Dataset¶

Get a Dataset object corresponding to a dataset on the Driverless AI server. If the dataset only exists on H2O.ai Storage, it will be imported to the server first.

Parameters:	key (`str`) – Driverless AI server’s unique ID for the dataset

Examples:

# Use the UUID of the dataset to retrieve the dataset object
key = 'e7de8630-dbfb-11ea-9f69-0242ac110002'
ds = dai.datasets.get(key=key)

Return type:	`Dataset`

gui() → Hyperlink¶

Get full URL for the user’s datasets page on Driverless AI server.

Examples:

dai.datasets.gui()

Return type:	`Hyperlink`

list(start_index: int = 0, count: Optional[int] = None) → Sequence[Dataset]¶

List of Dataset objects available to the user.

Parameters:	start_index (`int`) – index on Driverless AI server of first dataset to list count (`Optional`[`int`]) – max number of datasets to request from the Driverless AI server

Examples:

dai.datasets.list(start_index=10, count=5)

Return type:	`Sequence`[`Dataset`]

Experiments¶

API Reference:

class Experiments¶

Interact with experiments on the Driverless AI server.

create(train_dataset: Dataset, target_column: Optional[str], task: str, force: bool = False, name: Optional[str] = None, **kwargs: Any) → Experiment¶

Launch an experiment on the Driverless AI server and wait for it to complete before returning.

Parameters:

train_dataset (Dataset) – Dataset object
target_column (Optional[str]) – name of column in train_dataset (pass None if task is 'unsupervised')
task (str) – one of 'regression', 'classification', or 'unsupervised'
force (bool) – create new experiment even if experiment with same name already exists
name (Optional[str]) – display name for experiment

Keyword Arguments:

accuracy (int) – accuracy setting [1-10]
time (int) – time setting [1-10]
interpretability (int) – interpretability setting [1-10]
scorer (Union[str,ScorerRecipe]) – metric to optimize for
models (Union[str,ModelRecipe]) – limit experiment to these models
transformers (Union[str,TransformerRecipe]) – limit experiment to these transformers
validation_dataset (Dataset) – Dataset object
test_dataset (Dataset) – Dataset object
weight_column (str) – name of column in train_dataset
fold_column (str) – name of column in train_dataset
time_column (str) – name of column in train_dataset, containing time ordering for timeseries problems
time_groups_columns (List[str]) – list of column names, contributing to time ordering
unavailable_at_prediction_time_columns (List[str]) – list of column names, which won’t be present at prediction time
drop_columns (List[str]) – list of column names to be dropped
enable_gpus (bool) – allow GPU usage in experiment
reproducible (bool) – set experiment to be reproducible
time_period_in_seconds (int) – the length of the time period in seconds, used in timeseries problems
num_prediction_periods (int) – timeseries forecast horizon in time period units
num_gap_periods (int) – number of time periods after which forecast starts
config_overrides (str) – Driverless AI config overrides in TOML string format

Note

Any expert setting can also be passed as a kwarg. To search possible expert settings for your server version, use experiments.search_expert_settings(search_term).

Return type:	`Experiment`

create_async(train_dataset: Dataset, target_column: Optional[str], task: str, force: bool = False, name: Optional[str] = None, **kwargs: Any) → Experiment¶

Launch an experiment on the Driverless AI server and return an Experiment object to track the experiment status.

Parameters:

train_dataset (Dataset) – Dataset object
target_column (Optional[str]) – name of column in train_dataset (pass None if task is 'unsupervised')
task (str) – one of 'regression', 'classification', or 'unsupervised'
force (bool) – create new experiment even if experiment with same name already exists
name (Optional[str]) – display name for experiment

Keyword Arguments:

accuracy (int) – accuracy setting [1-10]
time (int) – time setting [1-10]
interpretability (int) – interpretability setting [1-10]
scorer (Union[str,ScorerRecipe]) – metric to optimize for
models (Union[str,ModelRecipe]) – limit experiment to these models
transformers (Union[str,TransformerRecipe]) – limit experiment to these transformers
validation_dataset (Dataset) – Dataset object
test_dataset (Dataset) – Dataset object
weight_column (str) – name of column in train_dataset
fold_column (str) – name of column in train_dataset
time_column (str) – name of column in train_dataset, containing time ordering for timeseries problems
time_groups_columns (List[str]) – list of column names, contributing to time ordering
unavailable_at_prediction_time_columns (List[str]) – list of column names, which won’t be present at prediction time
drop_columns (List[str]) – list of column names to be dropped
enable_gpus (bool) – allow GPU usage in experiment
reproducible (bool) – set experiment to be reproducible. If seed value set and reproducible was not set it will be marked as true by the client.
time_period_in_seconds (int) – the length of the time period in seconds, used in timeseries problems
num_prediction_periods (int) – timeseries forecast horizon in time period units
num_gap_periods (int) – number of time periods after which forecast starts
config_overrides (str) – Driverless AI config overrides in TOML string format

Note

Any expert setting can also be passed as a kwarg. To search possible expert settings for your server version, use experiments.search_expert_settings(search_term).

Return type:	`Experiment`

get(key: str) → Experiment¶

Get an Experiment object corresponding to an experiment on the Driverless AI server. If the experiment only exists on H2O.ai Storage, it will be imported to the server first.

Parameters:	key (`str`) – Driverless AI server’s unique ID for the experiment
Return type:	`Experiment`

gui() → Hyperlink¶

Get full URL for the experiments page on the Driverless AI server.

Return type:	`Hyperlink`

import_dai_file(path: str) → Experiment¶

Import a dai file to the Driverless AI server and return a: corresponding Experiment object.

Warning

Requires DriverlessAI server version 1.10.0 or higher.

Return type:	`Experiment`

leaderboard(train_dataset: Dataset, target_column: Optional[str], task: str, force: bool = False, name: str = None, **kwargs: Any) → Project¶

Launch an experiment leaderboard on the Driverless AI server and

return a project object to track experiment statuses. Requires server version >= 1.9.0.

Args:

train_dataset: Dataset object target_column: name of column in train_dataset

(pass None if task is 'unsupervised')

task: one of 'regression', 'classification', or 'unsupervised' force: create new project even if project with same name

already exists

name: display name for project

Keyword Args:

accuracy (int): accuracy setting [1-10] time (int): time setting [1-10] interpretability (int): interpretability setting [1-10] scorer (Union[str,ScorerRecipe]): metric to optimize for models (Union[str,ModelRecipe]): limit experiment to these models transformers (Union[str,TransformerRecipe]): limit experiment to

these transformers

validation_dataset (Dataset): Dataset object test_dataset (Dataset): Dataset object weight_column (str): name of column in train_dataset fold_column (str): name of column in train_dataset time_column (str): name of column in train_dataset,

containing time ordering for timeseries problems

time_groups_columns (List[str]): list of column names,: contributing to time ordering
unavailable_at_prediction_time_columns (List[str]):: list of column names, which won’t be present at prediction time

drop_columns (List[str]): list of column names to be dropped enable_gpus (bool): allow GPU usage in experiment reproducible (bool): set experiment to be reproducible time_period_in_seconds (int): the length of the time period in seconds,

used in timeseries problems

num_prediction_periods (int): timeseries forecast horizon in time: period units
num_gap_periods (int): number of time periods after which: forecast starts

config_overrides (str): Driverless AI config overrides in TOML string format

Note

Any expert setting can also be passed as a kwarg. To search possible expert settings for your server version, use experiments.search_expert_settings(search_term).

Warning

Requires DriverlessAI server version 1.9.0 or higher.

Return type:	`Project`

list(start_index: int = 0, count: Optional[int] = None) → Sequence[Experiment]¶

List of Experiment objects available to the user.

Parameters:	start_index (`int`) – index on Driverless AI server of first experiment in list count (`Optional`[`int`]) – number of experiments to request from the Driverless AI server
Return type:	`Sequence`[`Experiment`]

preview(train_dataset: Dataset, target_column: Optional[str], task: str, force: Optional[bool] = None, name: Optional[str] = None, **kwargs: Any) → None¶

Print a preview of experiment for the given settings.

Parameters:

train_dataset (Dataset) – Dataset object
target_column (Optional[str]) – name of column in train_dataset (pass None if task is 'unsupervised')
task (str) – one of 'regression', 'classification', or 'unsupervised'
force (Optional[bool]) – ignored (preview accepts the same arguments as create)
name (Optional[str]) – ignored (preview accepts the same arguments as create)

Keyword Arguments:

accuracy (int) – accuracy setting [1-10]
time (int) – time setting [1-10]
interpretability (int) – interpretability setting [1-10]
scorer (Union[str,ScorerRecipe]) – metric to optimize for
models (Union[str,ModelRecipe]) – limit experiment to these models
transformers (Union[str,TransformerRecipe]) – limit experiment to these transformers
validation_dataset (Dataset) – Dataset object
test_dataset (Dataset) – Dataset object
weight_column (str) – name of column in train_dataset
fold_column (str) – name of column in train_dataset
time_column (str) – name of column in train_dataset, containing time ordering for timeseries problems
time_groups_columns (List[str]) – list of column names, contributing to time ordering
unavailable_at_prediction_time_columns (List[str]) – list of column names, which won’t be present at prediction time
drop_columns (List[str]) – list of column names to be dropped
enable_gpus (bool) – allow GPU usage in experiment (True), to not allow GPU usage (False). If not set Driverless AI will decide whether CPU or GPU is best for the experiment
reproducible (bool) – set experiment to be reproducible
time_period_in_seconds (int) – the length of the time period in seconds, used in timeseries problems
num_prediction_periods (int) – timeseries forecast horizon in time period units
num_gap_periods (int) – number of time periods after which forecast starts
config_overrides (str) – Driverless AI config overrides in TOML string format

Note

Any expert setting can also be passed as a kwarg. To search possible expert settings for your server version, use experiments.search_expert_settings(search_term).

Return type:	`None`

search_expert_settings(search_term: str, show_description: bool = False) → None¶

Search expert settings and print results. Useful when looking for kwargs to use when creating experiments.

Parameters:	search_term (`str`) – term to search for (case-insensitive) show_description (`bool`) – include description in results
Return type:	`None`

MLI¶

API Reference:

class MLI¶

Interact with MLI interpretations on the Driverless AI server.

create(experiment: Optional[Experiment] = None, dataset: Optional[Dataset] = None, name: Optional[str] = None, force: bool = False, **kwargs: Any) → Interpretation¶

Create an MLI interpretation on the Driverless AI server and return an Interpretation object corresponding to the created interpretation.

Parameters:

experiment (Optional[Experiment]) – experiment to interpret, will use training dataset if dataset not specified
dataset (Optional[Dataset]) – dataset to use for interpretation (if dataset includes target and prediction columns, then an experiment is not needed)
name (Optional[str]) – display name for the interpretation
force (bool) – create new interpretation even if interpretation with same name already exists

Keyword Arguments:

explainers (List[ExplainerRecipe]) – list of explainer recipe objects
test_dataset (Dataset) – Dataset object (timeseries only)
target_column (str) – name of column in dataset
prediction_column (str) – name of column in dataset
weight_column (str) – name of column in dataset
drop_columns (List[str]) – names of columns in dataset

Note

Any expert setting can also be passed as a kwarg. To search possible expert settings for your server version, use mli.search_expert_settings(search_term).

Return type:	`Interpretation`

create_async(experiment: Optional[Experiment] = None, dataset: Optional[Dataset] = None, name: Optional[str] = None, force: bool = False, validate_settings: bool = True, **kwargs: Any) → Interpretation¶

Launch creation of an MLI interpretation on the Driverless AI server and return an Interpretation object to track the status.

Parameters:

experiment (Optional[Experiment]) – experiment to interpret, will use training dataset if dataset not specified
dataset (Optional[Dataset]) – dataset to use for interpretation (if dataset includes target and prediction columns, then an experiment is not needed)
name (Optional[str]) – display name for the interpretation
force (bool) – create new interpretation even if interpretation with same name already exists

Keyword Arguments:

explainers (List[ExplainerRecipe]) – list of explainer recipe objects (server versions >= 1.9.1)
test_dataset (Dataset) – Dataset object (timeseries only)
target_column (str) – name of column in dataset
prediction_column (str) – name of column in dataset
weight_column (str) – name of column in dataset
drop_columns (List[str]) – names of columns in dataset

Note

Any expert setting can also be passed as a kwarg. To search possible expert settings for your server version, use mli.search_expert_settings(search_term).

Return type:	`Interpretation`

gui() → Hyperlink¶

Print full URL for the user’s MLI page on Driverless AI server.

Return type:	`Hyperlink`

property iid: driverlessai._mli.IIDMethods¶

Retrieve IID interpretations.

Return type:	`IIDMethods`

search_expert_settings(search_term: str = '', show_description: bool = False, show_valid_values: bool = False) → Table¶

Search expert settings and print results. Useful when looking for kwargs to use when creating interpretations.

Parameters:	search_term (`str`) – term to search for (case-insensitive) show_description (`bool`) – include description in results show_valid_values (`bool`) – include valid values that can be set for each setting in the results
Return type:	`Table`

property timeseries: driverlessai._mli.TimeseriesMethods¶

Retrieve timeseries interpretations.

Return type:	`TimeseriesMethods`

Interpretation Retrieval Methods¶

class IIDMethods¶

get(key: str) → Interpretation¶

Get an Interpretation object corresponding to an MLI interpretation on the Driverless AI server.

Parameters:	key (`str`) – Driverless AI server’s unique ID for the MLI interpretation
Return type:	`Interpretation`

list(start_index: int = 0, count: Optional[int] = None) → Sequence[Interpretation]¶

List of Interpretation objects available to the user.

Parameters:	start_index (`int`) – index on Driverless AI server of first interpretation to list count (`Optional`[`int`]) – max number of interpretations to request from the Driverless AI server
Return type:	`Sequence`[`Interpretation`]

class TimeseriesMethods¶

get(key: str) → Interpretation¶

Get an Interpretation object corresponding to an MLI interpretation on the Driverless AI server.

Parameters:	key (`str`) – Driverless AI server’s unique ID for the MLI interpretation
Return type:	`Interpretation`

list(start_index: int = 0, count: Optional[int] = None) → Sequence[Interpretation]¶

List of Interpretation objects available to the user.

Parameters:	start_index (`int`) – index on Driverless AI server of first interpretation to list count (`Optional`[`int`]) – max number of interpretations to request from the Driverless AI server
Return type:	`Sequence`[`Interpretation`]

Model Diagnostics¶

API Reference:

class ModelDiagnostics¶

Interact with model diagnostics on the Driverless AI server.

create(diagnose_experiment: Experiment, test_dataset: Dataset) → ModelDiagnostic¶

Create a new model diagnostic.

Parameters:	diagnose_experiment (`Experiment`) – experiment that created the diagnosing model test_dataset (`Dataset`) – test dataset for the diagnosis
Return type:	`ModelDiagnostic`

create_async(diagnose_experiment: Experiment, test_dataset: Dataset) → ModelDiagnosticJob¶

Launch creation of a new model diagnostic.

Parameters:	diagnose_experiment (`Experiment`) – experiment that created the diagnosing model test_dataset (`Dataset`) – test dataset for the diagnosis
Return type:	`ModelDiagnosticJob`

get(key: str) → ModelDiagnostic¶

Get a ModelDiagnostic object corresponding to a model diagnostic on the Driverless AI server.

Parameters:	key (`str`) – Driverless AI server’s unique ID for the model diagnostic
Return type:	`ModelDiagnostic`

gui() → Hyperlink¶

Get full URL for the Model Diagnostics page on the Driverless AI server.

Return type:	`Hyperlink`

list(start_index: int = 0, count: Optional[int] = None) → Sequence[ModelDiagnostic]¶

Return list of ModelDiagnostic objects on the Driverless AI server.

Parameters:	start_index (`int`) – index on Driverless AI server of first model diagnostic in list count (`Optional`[`int`]) – number of model diagnostics to request from the Driverless AI server
Return type:	`Sequence`[`ModelDiagnostic`]

class ModelDiagnostic¶

Interact with a model diagnostic on the Driverless AI server.

delete() → None¶

Delete this model diagnostic on Driverless AI server.

Return type:	`None`

download_predictions(dst_dir: str = '.', dst_file: Optional[str] = None, file_system: Optional[fsspec.spec.AbstractFileSystem] = None, overwrite: bool = False, timeout: float = 30) → str¶

Downloads the predictions into a csv file.

Return type:	`str`

property experiment: driverlessai._experiments.Experiment¶

Diagnosed experiment by this model diagnostic.

Return type:	`Experiment`

gui() → Hyperlink¶

Get full URL for this model diagnostic’s page on the Driverless AI server.

Return type:	`Hyperlink`

property key: str¶

Universally unique identifier.

Return type:	`str`

property metric_plots: driverlessai._model_diagnostics.ModelDiagnosticMetricPlots¶

Metric plots of this model diagnostic.

Warning

Requires DriverlessAI server version 1.9.0 or higher.

Warning

A beta API that is subject to future changes.

Return type:	`ModelDiagnosticMetricPlots`

property name: str¶

Display name.

Return type:	`str`

property scores: Dict[str, Dict[str, float]]¶

Scores of this model diagnostic.

Return type:	`Dict`[`str`, `Dict`[`str`, `float`]]

property test_dataset: driverlessai._datasets.Dataset¶

Test dataset that was used for this model diagnostic.

Return type:	`Dataset`

class ModelDiagnosticMetricPlots¶

Interacts with metric plots of a model diagnostic on the Driverless AI server.

property actual_vs_predicted_chart: Optional[Dict[str, Any]]¶

Actual vs predicted Chart of this model diagnostic.

Return type:	`Optional`[`Dict`[`str`, `Any`]]
Returns:	an actual vs predicted chart in Vega Lite (v3) format

confusion_matrix(threshold: Optional[float] = None) → Optional[List[List[Any]]]¶

Confusion Matrix of this model diagnostic.

Parameters:	threshold (`Optional`[`float`]) – a threshold value
Return type:	`Optional`[`List`[`List`[`Any`]]]
Returns:	the confusion matrix as a 2D list

property gains_chart: Optional[Dict[str, Any]]¶

Cumulative Gain Chart of this model diagnostic.

Return type:	`Optional`[`Dict`[`str`, `Any`]]
Returns:	a gains chart in Vega Lite (v3) format

property ks_chart: Optional[Dict[str, Any]]¶

Kolmogorov-Smirnov Chart of this model diagnostic.

Return type:	`Optional`[`Dict`[`str`, `Any`]]
Returns:	a Kolmogorov-Smirnov chart in Vega Lite (v3) format

property lift_chart: Optional[Dict[str, Any]]¶

Lift Chart of this model diagnostic.

Return type:	`Optional`[`Dict`[`str`, `Any`]]
Returns:	a lift chart in Vega Lite (v3) format

property prec_recall_curve: Optional[Dict[str, Any]]¶

Precision-Recall Curve of this model diagnostic.

Return type:	`Optional`[`Dict`[`str`, `Any`]]
Returns:	a precision-recall chart in Vega Lite (v3) format

property residual_histogram: Optional[Dict[str, Any]]¶

Residual Histogram of this model diagnostic.

Return type:	`Optional`[`Dict`[`str`, `Any`]]
Returns:	a residual histogram in Vega Lite (v3) format

property residual_plot: Optional[Dict[str, Any]]¶

Residual Plot with LOESS Curve of this model diagnostic.

Return type:	`Optional`[`Dict`[`str`, `Any`]]
Returns:	a residual plot in Vega Lite (v3) format

property roc_curve: Optional[Dict[str, Any]]¶

ROC Curve of this model diagnostic.

Return type:	`Optional`[`Dict`[`str`, `Any`]]
Returns:	a ROC curve in Vega Lite (v3) format

Projects¶

API Reference:

class Projects¶

Interact with projects on the Driverless AI server.

create(name: str, description: Optional[str] = None, force: bool = False) → Project¶

Create a project on the Driverless AI server.

Parameters:	name (`str`) – display name for project description (`Optional`[`str`]) – description of project force (`bool`) – create new project even if project with same name already exists
Return type:	`Project`

get(key: str) → Project¶

Get a Project object corresponding to a project on the Driverless AI server.

Parameters:	key (`str`) – Driverless AI server’s unique ID for the project
Return type:	`Project`

gui() → Hyperlink¶

Get full URL for the projects page on the Driverless AI server.

Return type:	`Hyperlink`

list(start_index: int = 0, count: Optional[int] = None) → Sequence[Project]¶

List of Project objects available to the user.

Parameters:	start_index (`int`) – index on Driverless AI server of first project in list count (`Optional`[`int`]) – number of projects to request from the Driverless AI server
Return type:	`Sequence`[`Project`]

Recipes¶

API Reference:

class Recipes¶

Create and interact with recipes on the Driverless AI server.

create(recipe: str, username: Optional[str] = None, password: Optional[str] = None) → Recipe¶

Create a recipe on the Driverless AI server.

Parameters:	recipe (`str`) – path or URL to the recipe username (`Optional`[`str`]) – username to use when connecting to BitBucket password (`Optional`[`str`]) – password to use when connecting to BitBucket

Examples:

dai.recipes.create(
    recipe='https://github.com/h2oai/driverlessai-recipes/blob/master/scorers/regression/explained_variance.py'
)

Return type:	`Recipe`

create_async(recipe: str, username: Optional[str] = None, password: Optional[str] = None) → RecipeJob¶

Launch creation of a recipe on the Driverless AI server.

Parameters:	recipe (`str`) – path or URL to the recipe username (`Optional`[`str`]) – username to use when connecting to BitBucket password (`Optional`[`str`]) – password to use when connecting to BitBucket

Examples:

dai.recipes.create_async(
    recipe='https://github.com/h2oai/driverlessai-recipes/blob/master/scorers/regression/explained_variance.py'
)

Return type:	`RecipeJob`

property data: driverlessai._recipes.DataRecipes¶

See data recipes on the Driverless AI server.

Warning

Requires DriverlessAI server version 1.10.0 or higher.

Return type:	`DataRecipes`

property explainers: driverlessai._recipes.ExplainerRecipes¶

See explainer recipes on the Driverless AI server.

Warning

Requires DriverlessAI server version 1.9.1 or higher.

Return type:	`ExplainerRecipes`

property individuals: driverlessai._recipes.IndividualRecipes¶

See individual recipes on the Driverless AI server.

Warning

Requires DriverlessAI server version 1.10.2 or higher.

Return type:	`IndividualRecipes`

property models: driverlessai._recipes.ModelRecipes¶

See model recipes on the Driverless AI server.

Return type:	`ModelRecipes`

property pre_transformers: driverlessai._recipes.PreTransformerRecipes¶

See pre-transformer recipes on the Driverless AI server.

Warning

Requires DriverlessAI server version 1.9.0 or higher.

Return type:	`PreTransformerRecipes`

property scorers: driverlessai._recipes.ScorerRecipes¶

See scorer recipes on the Driverless AI server.

Return type:	`ScorerRecipes`

property transformers: driverlessai._recipes.TransformerRecipes¶

See transformer recipes on the Driverless AI server.

Return type:	`TransformerRecipes`

class ExplainerRecipes¶

Interact with explainer recipes on the Driverless AI server.

Examples:

# Get list of names of all explainers
[e.name for e in dai.recipes.explainers.list()]

list() → Sequence[ExplainerRecipe]¶

Return list of explainer recipe objects.

Examples:

dai.recipes.explainer.list()

Return type:	`Sequence`[`ExplainerRecipe`]

class ModelRecipes¶

Interact with model recipes on the Driverless AI server.

Examples:

# Get list of all custom models
[m for m in dai.recipes.models.list() if m.is_custom]

# Get list of names of all models
[m.name for m in dai.recipes.models.list()]

list() → Sequence[ModelRecipe]¶

Return list of model recipe objects.

Examples:

dai.recipes.models.list()

Return type:	`Sequence`[`ModelRecipe`]

class ScorerRecipes¶

Interact with scorer recipes on the Driverless AI server.

Examples:

# Retrieve a list of binomial scorers
[s for s in dai.recipes.scorers.list() if s.for_binomial]

# Retrieve a list of multiclass scorers
[s for s in dai.recipes.scorers.list() if s.for_multiclass]

# Retrieve a list of regression scorers
[s for s in dai.recipes.scorers.list() if s.for_regression]

# Get list of all custom scorers
[s for s in dai.recipes.scorers.list() if s.is_custom]

# Get list of names of all scorers
[s.name for s in dai.recipes.scorers.list()]

# Get list of descriptions for all scorers
[s.description for s in dai.recipes.scorers.list()]

list() → Sequence[ScorerRecipe]¶

Return list of scorer recipe objects.

Examples:

dai.recipes.scorers.list()

Return type:	`Sequence`[`ScorerRecipe`]

class TransformerRecipes¶

Interact with transformer recipes on the Driverless AI server.

Examples:

# Get list of all custom transformers
[m for m in dai.recipes.transformers.list() if m.is_custom]

# Get list of names of all transformers
[m.name for m in dai.recipes.transformers.list()]

list() → Sequence[TransformerRecipe]¶

Return list of transformer recipe objects.

Examples:

dai.recipes.transformers.list()

Return type:	`Sequence`[`TransformerRecipe`]

Get Information about the Server¶

The following attribute of a connected client allows you to get information about the Driverless AI server.

Server¶

API Reference:

class Server¶

Get information about the Driverless AI server.

Examples:

# Connect to the DAI server
dai = driverlessai.Client(
    address='http://localhost:12345',
    username='py',
    password='py'
)

dai.server.address
dai.server.username
dai.server.version

property address: str¶

URL of the Driverless AI server connected to.

Return type:	`str`

property cpu_usages: List[float]¶

Get usage of each CPU of the Driverless AI server.

Warning

Requires DriverlessAI server version 1.10.6 or higher.

Warning

A beta API that is subject to future changes.

Return type:	`List`[`float`]

property disk_usage: Dict[str, int]¶

Get disk usage(in bytes) of the Driverless AI server.

Return type:	`Dict`[`str`, `int`]

docs(search: Optional[str] = None) → Hyperlink¶

Get link to documentation on the Driverless AI server.

Parameters:	search (`Optional`[`str`]) – if search terms are supplied, the link will go to documentation search results

Example:

# Search the DAI docs for "licenses"
dai.server.docs(search='licenses')

Return type:	`Hyperlink`

property experiment_stats: Dict[str, int]¶

Get experiments related statistics of the Driverless AI server.

Return type:	`Dict`[`str`, `int`]

property gpu_usages: List[Dict[str, float]]¶

Get GPU statistics of the Driverless AI server.

Return type:	`List`[`Dict`[`str`, `float`]]

gui() → Hyperlink¶

Get full URL for the Driverless AI server.

Examples:

dai.server.gui()

Return type:	`Hyperlink`

property license: driverlessai._server.License¶

Get information about the license on the Driverless AI server.

Return type:	`License`

property memory_usage: Dict[str, int]¶

Get memory statistics(in bytes) of the Driverless AI server.

Warning

Requires DriverlessAI server version 1.10.6 or higher.

Return type:	`Dict`[`str`, `int`]

property storage_enabled: bool¶

Whether the Driverless AI server is connected to H2O.ai Storage.

Return type:	`bool`

property username: str¶

Current user connected as to a Driverless AI server.

Return type:	`str`

property version: str¶

Version of Driverless AI server currently connected to.

Return type:	`str`

property workers: List[Dict[str, Any]]¶

Get statistics each worker of the Driverless AI server.

Warning

Requires DriverlessAI server version 1.9.0.6 or higher.

Warning

A beta API that is subject to future changes.

Return type:	`List`[`Dict`[`str`, `Any`]]

class License¶

Get information about the Driverless AI server’s license.

days_left() → int¶

Return days left on license.

Examples:

dai.server.license.days_left()

Return type:	`int`

is_valid() → bool¶

Return True if server is licensed.

Examples:

dai.server.license.is_valid()

Return type:	`bool`