Prediction results

This schema defines an intermediate artifact from sprint prediction.

Sprint labels

https://gros.liacs.nl/schema/prediction/sprint_labels.json
Results of predictions.
oneOf	sprint_labels
	type	array
	items	sprint_labels

sprint_labels

type	object
properties
labels	type	array
	items	Classification label for the sprint from the model.
		type	integer
probabilities	type	array
	items	Reliability of the predicted label for the sprint according to the estimator model.
		type	number
		maximum	1.0
		minimum	0.0
risks	type	array
	items	Chance that the sprint will receive a negative classification after all, based on internal activations of the estimator model but with external normalized scale.
		type	number
		maximum	1.0
		minimum	0.0
analogy_distances	type	array
	items	Distances to the closest sprint samples based on a normalized feature distance metric.
		type	array
		items	type	number
analogy_indexes	type	array
	items	Array indexes of the closest sprint samples.
		type	array
		items	type	integer
analogy_labels	type	array
	items	Labels of the closest sprint samples.
		type	array
		items	type	integer
analogy_values	type	array
	items	Values of the features of the closest sprint samples.
		type	array
		items	type	array
			items	type	number / null
metrics	type	object
	patternProperties
	.*	type	number
projects	type	array
	items	Identifier of the project which the sprint that is being labeled at this index is from.
		type	integer
sprints	type	array
	items	Identifier of the sprint that is being labeled at this index.
		type	integer
organizations	type	array
	items	Organization which the sprint that is being labeled at this index is from.
		type	string
features	type	array
	items	Values of the features for a sample sprint, used as input for the model.
		type	array
		items	type	number / null
configuration	type	object
	properties
	label	Name or expression of the label used as target output in the model.
		type	string
	labels	type	array
		items	Name of a feature used as part of the expression for the target output label.
			type	string
	assignments	type	object
		patternProperties
		.*	type	array
			items	Feature used in an assignment of another generated feature.
				type	string
	features	type	array
		items	Name of a feature used as input for the model.
			type	string
	metadata	type	array
		items	Name of a column of the data set used as metadata.
			type	string
	model	Identifier of the model used.
		type	string
	binary	Treshold at which a label value is converted to a binary classification, if provided.
		type	number / null
	weighted	Whether input values were weighted to produce values at similar scales as input for the model.
		type	boolean
	stratified	Whether the data set was resampled (some sample sprints provided more often) in order to create proportionally balanced subsets with normal distributions of the target label, when provided in batches to the model.
		type	boolean