Competition

polaris.competition.CompetitionSpecification

Bases: DatasetV2, PredictiveTaskSpecificationMixin, SplitSpecificationV1Mixin

An instance of this class represents a Polaris competition. It defines fields and functionality that in combination with the DatasetV2 class, allow users to participate in competitions hosted on Polaris Hub.

Examples:

Basic API usage:

import polaris as po

# Load the benchmark from the Hub
competition = po.load_competition("dummy-user/dummy-name")

# Get the train and test data-loaders
train, test = competition.get_train_test_split()

# Use the training data to train your model
# Get the input as an array with 'train.inputs' and 'train.targets'
# Or simply iterate over the train object.
for x, y in train:
    ...

# Work your magic to accurately predict the test set
prediction_values = np.array([0.0 for x in test])

# Submit your predictions
competition.submit_predictions(
    prediction_name="first-prediction",
    prediction_owner="dummy-user",
    report_url="REPORT_URL",
    predictions=prediction_values,
)

Attributes:

Name	Type	Description
start_time	datetime	The time at which the competition starts accepting prediction submissions.
end_time	datetime	The time at which the competition stops accepting prediction submissions.
n_classes	dict[ColumnName, int | None]	The number of classes within each target column that defines a classification task.

For additional meta-data attributes, see the base classes.

get_train_test_split

get_train_test_split(
    featurization_fn: Callable | None = None,
) -> tuple[Subset, Subset | dict[str, Subset]]

Construct the train and test sets, given the split in the competition specification.

Returns Subset objects, which offer several ways of accessing the data and can thus easily serve as a basis to build framework-specific (e.g. PyTorch, Tensorflow) data-loaders on top of.

Parameters:

Name	Type	Description	Default
featurization_fn	Callable | None	A function to apply to the input data. If a multi-input benchmark, this function expects an input in the format specified by the input_format parameter.	None

Returns:

Type	Description
tuple[Subset, Subset | dict[str, Subset]]	A tuple with the train Subset and test Subset objects. If there are multiple test sets, these are returned in a dictionary and each test set has an associated name. The targets of the test set can not be accessed.

submit_predictions

submit_predictions(
    predictions: IncomingPredictionsType,
    prediction_name: SlugCompatibleStringType,
    prediction_owner: str,
    report_url: HttpUrlString,
    contributors: list[HubUser] | None = None,
    github_url: HttpUrlString | None = None,
    description: str = "",
    tags: list[str] | None = None,
    user_attributes: dict[str, str] | None = None,
) -> None

Convenient wrapper around the PolarisHubClient.submit_competition_predictions method. It handles the creation of a standardized predictions object, which is expected by the Hub, automatically.

Parameters:

Name	Type	Description	Default
prediction_name	SlugCompatibleStringType	The name of the prediction.	required
prediction_owner	str	The slug of the user/organization which owns the prediction.	required
predictions	IncomingPredictionsType	The predictions for each test set defined in the competition.	required
report_url	HttpUrlString	A URL to a report/paper/write-up which describes the methods used to generate the predictions.	required
contributors	list[HubUser] | None	The users credited with generating these predictions.	None
github_url	HttpUrlString | None	An optional URL to a code repository containing the code used to generated these predictions.	None
description	str	An optional and short description of the predictions.	''
tags	list[str] | None	An optional list of tags to categorize the prediction by.	None
user_attributes	dict[str, str] | None	An optional dict with additional, textual user attributes.	None

---