Filter observations to have a consistent maximum delay period
Source:R/preprocess.R
enw_filter_delay.RdFilter observations to have a consistent maximum delay period
Arguments
- obs
A
data.framecontaining at least the following variables:reference date(index date of interest),report_date(report date for observations), andconfirm(cumulative observations by reference and report date).- max_delay
The maximum delay to model in the delay distribution, specified in units of the timestep (e.g., if
timestep = "week", thenmax_delay = 3means 3 weeks). If not specified the maximum observed delay is assumed to be the true maximum delay in the model. Otherwise, an integer greater than or equal to 1 can be specified. Observations with delays larger than the maximum delay will be dropped. If the specified maximum delay is too short, nowcasts can be biased as important parts of the true delay distribution are cut off. At the same time, computational cost scales non-linearly with this setting, so you want the maximum delay to be as long as necessary, but not much longer.Steps to take to determine the maximum delay:
Consider what is realistic and relevant for your application.
Check the proportion of observations reported (
prop_reported) by delay in thenew_confirmoutput ofenw_preprocess_obs.Use
check_max_delay()to check the coverage of a candidatemax_delay.If in doubt, check if increasing the maximum delay noticeably changes the delay distribution or nowcasts as estimated by
epinowcast. If it does, your maximum delay may still be too short.
Note that delays are zero indexed and so include the reference date and
max_delay - 1other intervals (i.e. amax_delayof 1 corresponds to no delay).- timestep
The timestep to used in the process model (i.e. the reference date model). This can be a string ("day", "week", "month") or a numeric whole number representing the number of days. If your data does not have this timestep then you may wish to make use of
enw_aggregate_cumulative()to aggregate your data to the desired timestep.
Value
A data.frame filtered so that dates by report are less than or
equal the reference date plus the maximum delay.
See also
Preprocessing functions
enw_add_delay(),
enw_add_max_reported(),
enw_add_metaobs_features(),
enw_assign_group(),
enw_complete_dates(),
enw_construct_data(),
enw_extend_date(),
enw_filter_reference_dates(),
enw_filter_report_dates(),
enw_flag_observed_observations(),
enw_impute_na_observations(),
enw_latest_data(),
enw_metadata(),
enw_metadata_delay(),
enw_missing_reference(),
enw_preprocess_data(),
enw_reporting_triangle(),
enw_reporting_triangle_to_long()
Examples
obs <- enw_example("preprocessed")$obs[[1]]
enw_filter_delay(obs, max_delay = 2)
#> reference_date .group report_date max_confirm location age_group confirm
#> <IDat> <num> <IDat> <int> <fctr> <fctr> <int>
#> 1: <NA> 1 2021-07-14 0 DE 00+ 0
#> 2: <NA> 1 2021-07-15 0 DE 00+ 0
#> 3: <NA> 1 2021-07-16 0 DE 00+ 0
#> 4: <NA> 1 2021-07-17 0 DE 00+ 0
#> 5: <NA> 1 2021-07-18 0 DE 00+ 0
#> ---
#> 115: 2021-08-20 1 2021-08-20 171 DE 00+ 98
#> 116: 2021-08-20 1 2021-08-21 171 DE 00+ 159
#> 117: 2021-08-21 1 2021-08-21 112 DE 00+ 69
#> 118: 2021-08-21 1 2021-08-22 112 DE 00+ 112
#> 119: 2021-08-22 1 2021-08-22 45 DE 00+ 45
#> cum_prop_reported delay
#> <num> <num>
#> 1: NaN NA
#> 2: NaN NA
#> 3: NaN NA
#> 4: NaN NA
#> 5: NaN NA
#> ---
#> 115: 0.5730994 0
#> 116: 0.9298246 1
#> 117: 0.6160714 0
#> 118: 1.0000000 1
#> 119: 1.0000000 0