factor_agent.acceptance.AcceptanceReport

class AcceptanceReport(segment, train_window, test_window, seeds, top_k, accepted, baseline, with_accepted, selection=False, forced=False, holdout_opens=(), code_version=<factory>, contributions=())

Bases: object

The baseline against the baseline plus the accepted factors, seed by seed and on average.

accepted maps each factor added to the baseline, by label, to its output name in the model: every accepted factor, or the members of a selection when selection is true. forced is true when the hold-out was opened past its lock; holdout_opens lists every open of the hold-out so far (empty for a validation report). contributions holds each accepted factor’s contribution per mode (acceptance.contribution), empty when none was asked for; there is no threshold.

Methods

AcceptanceReport.contribution_table	One row per factor and mode: factor, mode, difference and agreeing of every statistic.
AcceptanceReport.markdown	The report as a short Markdown page.
AcceptanceReport.to_dict	The report as <segment>.json holds it, with the seed means, the difference and the contribution table.

Attributes

AcceptanceReport.contributions = ()

AcceptanceReport.difference

With the accepted factors minus the baseline, of each seed-averaged statistic.

AcceptanceReport.forced = False

AcceptanceReport.holdout_opens = ()

AcceptanceReport.selection = False

AcceptanceReport.segment

AcceptanceReport.train_window

AcceptanceReport.test_window

AcceptanceReport.seeds

AcceptanceReport.top_k

AcceptanceReport.accepted

AcceptanceReport.baseline

AcceptanceReport.with_accepted

AcceptanceReport.code_version