FlexMeasures / flexmeasures / 37767401158

87%

push

github

fix: count every missing row towards the forecaster's missing threshold, and fill forecast input gaps in pandas (#2640)

* perf: fill forecast input gaps in pandas instead of darts

Building a Darts TimeSeries and running gaps() up to three times per forecast step was
most of a forecast's input preparation. Same results, including the gap-row count.

Signed-off-by: Stijn van Houwelingen <teadrinkingprogrammer@proton.me>

* docs/changelog: link PR #2640 in the forecasting entry

Signed-off-by: Stijn van Houwelingen <teadrinkingprogrammer@proton.me>

* Count every missing row towards the forecaster's missing threshold

A gap counted one row fewer than it spans, so a single missing row between two readings counted as nothing,
and data missing every other row passed any threshold.
Missing rows are now counted at the target's resolution, a reading covering the rows its own sensor's resolution spans,
which also stops a regressor with a zero resolution from dividing by zero.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01B1NrAQfNXGbXNMoycCvorh
Signed-off-by: F.N. Claessen <claessen@seita.nl>

---------

Signed-off-by: Stijn van Houwelingen <teadrinkingprogrammer@proton.me>
Signed-off-by: F.N. Claessen <claessen@seita.nl>
Co-authored-by: F.N. Claessen <claessen@seita.nl>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

35 of 37 new or added lines in 1 file covered. (94.59%)

21209 of 24523 relevant lines covered (86.49%)

0.86 hits per line