00:00:00

hampel proposal 1

Review this independently proposed patch against the supplied starter. Trailing windows include both time endpoints and every observation tied at the anchor time, separately for each group. Trace the native model, algorithm services, repository snapshot and HTTP preview. Repair the incident while preserving the other contracts. All explicit fixtures and mandatory regression groups are available with bash test/run.sh.

Review focus: robust-statistics

src/services/windows.py

BeforeProposed

+1−1

33def windows(frame,width):

44 left=frame.alias('a')

55 right=frame.alias('s')

6 condition=(F.col('a.group')==F.col('s.group')) & (F.col('s.time')>F.col('a.time')-width) & (F.col('s.time')<=F.col('a.time'))

6 condition=(F.col('a.group')==F.col('s.group')) & (F.col('s.time')>=F.col('a.time')-width) & (F.col('s.time')<=F.col('a.time'))

77 return left.join(right,condition).select(F.col('a.id').alias('anchor'),F.col('s.value').alias('value'),F.col('s.weight').alias('weight'))

Your review

0 / 10,000 characters · A verdict needs supporting evidence.

30 more characters to give a verdict