Skip to content

group_times misses nearest hour with mins threshold #5

@robitalec

Description

@robitalec

If group_times threshold is minutes (e.g. 5 minutes), we aren't grouping to the nearest hour if we are nearer to next hour than e.g. 55 minutes.

Example:

  DT <- data.table(
  idate = as.IDate('2018-12-12'),
  itime = as.ITime(
    c('01:00', '01:02', '00:58',
      '03:00', '03:02', '02:59',
      '05:00', '05:01', '04:58')
  )
)
group_times(DT, c('idate', 'itime'), threshold = '5 minutes')
idate itime minutes timegroup
2018-12-12 01:00:00 0 1
2018-12-12 01:02:00 0 1
2018-12-12 00:58:00 60 2
2018-12-12 03:00:00 0 3
2018-12-12 03:02:00 0 3
2018-12-12 02:59:00 60 4
2018-12-12 05:00:00 0 5
2018-12-12 05:01:00 0 5
2018-12-12 04:58:00 60 6

We'd expect here to have 3 timegroups (on 1:00, 3:00 and 5:00).
We end up with 6.

Note: this kind of rolling to the nearest hour was addressed for the less frequent, rolling 24 hours to the next day.

Regardless, still miss grouping row 3 and row 9.

DT <- data.table(
  idate = as.IDate(
    c('2018-12-12','2018-12-12', '2018-12-12',
      '2018-12-12','2018-12-12', '2018-12-13',
      '2018-12-13','2018-12-13', '2018-12-13')),
    itime = as.ITime(
      c('22:00', '22:02', '21:58',
        '23:58', '23:59', '0:01',
        '2:00', '2:01', '1:58'))
)
  
group_times(DT, c('idate', 'itime'), threshold = '5 minutes')
idate itime minutes timegroup
2018-12-12 22:00:00 0 1
2018-12-12 22:02:00 0 1
2018-12-12 21:58:00 60 2
2018-12-12 23:58:00 60 3
2018-12-12 23:59:00 60 3
2018-12-13 00:01:00 0 3
2018-12-13 02:00:00 0 4
2018-12-13 02:01:00 0 4
2018-12-13 01:58:00 60 5

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions