revscoring.scoring

Scoring is what the revscoring library was designed to do. The basics of scoring are revscoring.Model that implement score() and revscoring.scoring.Statistics that are fit() using the scores generated by a revscoring.Model. Prediction models are fragile, so models keep track of their revscoring.scoring.Environment and you can revscoring.scoring.Environment.check() them against the current environment.

See revscoring.scoring.models and revscoring.scoring.statistics for more information.

class revscoring.Model(features, version=None, environment=None, statistics=None, additional_info=None)

dump(f)

Writes serialized model information to a file.

info = None

A revscoring.scoring.ModelInfo instance that implements lookup() and format() – both of which act as an index into information about a model.

classmethod load(f, error_on_env_check=False)

Reads serialized model information from a file.

score(feature_values)

Make a prediction or otherwise use the model to generate a score.

Parameters:

feature_values : collection(mixed)

an ordered collection of values that correspond to the Feature s provided to the constructor

Returns:

A dict of statistics

test(values_labels)

Tests the model against a labeled data.

Parameters:

values_labels : iterable (( <feature_values>, <label> ))

an iterable of labeled data Where <values_labels> is an ordered collection of predictive values that correspond to the Feature s provided to the constructor

Returns:

A dictionary of test results.

class revscoring.scoring.ModelInfo(pairs=[], default_fields=None)

format(paths=None, formatting='str', **kwargs)

Format a representation of the model information in a useful way.

Parameters:

paths : iterable ( str | [str] )

A set of paths to use when selecting which information should formatted. Everything beneath a provided path in the tree will be formatted. E.g. statistics.roc_auc and statistics will format redundantly because roc_auc is already within statistics. Alternatively statistics.roc_auc and statistics.pr_auc will format only those two specific bits of information.

formatting : “json” or “str”

Which output formatting do you want? “str” returns something nice to show on the command-line. “json” returns something that will pass through json.dump() without error.

lookup(path=None)

Looks up a specific information value based on either a string pattern or a path.

For example, the pattern “stats.roc_auc.labels.true” is the same as the path ['stats', 'roc_auc', 'labels', True].

Parameters:

path : str | list

The location of the information to lookup.