Skip to content

create_periods_t: bugfix, make last period same length as others - #33

Merged
mmyrte merged 3 commits into
mainfrom
bugfix/regular-period-lengths
Jul 31, 2026
Merged

create_periods_t: bugfix, make last period same length as others#33
mmyrte merged 3 commits into
mainfrom
bugfix/regular-period-lengths

Conversation

@mmyrte

@mmyrte mmyrte commented Jul 30, 2026

Copy link
Copy Markdown
Member

Periods need to be of equal length; now decided to err on the side of "create a last full period that may go beyond the specified end_extrapolated"

mmyrte and others added 3 commits July 30, 2026 11:12
The join in fit_partial_models() and fit_full_models() kept every row of
trans_preds_nested, so restricting trans_meta had no effect and non-viable
transitions could still be fitted with NA lulc ids. Make it an inner join.
Regression from e5df256.

Direct-learner mode has carried the no.crossval sentinel in crossval_score
since 39ea13a, so that predict_trans_pot() can rank the model; the tests and
the roxygen block still claimed it would be empty.

Regenerate R/sysdata.rda so test_periods_t reflects the create_periods_t fix
(its last period was still one day long), and guard the fixture against
future drift, since data-raw/ does not run during checks.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011DgJshsVtPMqdw7Ni7oKba
@mmyrte
mmyrte merged commit 35afe98 into main Jul 31, 2026
2 checks passed
@mmyrte
mmyrte deleted the bugfix/regular-period-lengths branch July 31, 2026 13:43
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants