Module detail 

What a test said message by message (sioul spam eval --errors, sioul spam train --errors, and the MCP’s spam_eval and spam_train): the test messages counted by account and by folder, their label (spam or ham, as decided) against the filter’s class (probably spam, maybe spam, probably not spam); the numbers at a grid of thresholds, so that one can be chosen; and the worst errors: ham the filter called spam or maybe spam, the surest first, then spam it did not call spam, the least sure first.

An error is said by where it is and by its card, as the Porch reads the message (its sender, its subject, the start of its text): the command line prints its sender and its subject, masked as the MCP masks them (codes, sign-in links, account numbers), never its text. Nothing here is written to a file: trained.toml keeps numbers only.

Structs

Confusion

Messages by their label, then by the filter’s class.

Detail

What a test said message by message.

Errors

The worst errors of a test set.

OutsideDetail

An outside source’s held-out part, message by message.

Row

How many messages of one label the filter judged probably spam, maybe spam, probably not spam.

Tested

One test message, as the filter judged it.

Wrong

One error: a ham the filter called spam or maybe spam, or a spam it did not call spam.

Enums

Origin

Where a test message comes from.

Constants

GRID

The thresholds the grid gives numbers at.

MOST_ERRORS

Errors listed at most, of each kind.

UNSETTLED_DAYS

Days after which a message left in an inbox counts as said ham: before, it may be spam nobody has looked at yet.

Functions

class_of

The class a probability falls in, between the two thresholds.

evidence_name

What a label’s evidence is called in the errors listed.

of

The detail of a test: counts, the grid, and up to errors errors of each kind for your own mail and for each outside source. The errors’ cards are read again from the corpus and the outside material, once.

unsettled

Whether a message’s label may not be settled yet: ham in an inbox, come less than UNSETTLED_DAYS before now.