Open
Description
Need to make miss_var_run and miss_var_span work for:
- multiple time series (mts), a facet for each variable.
- regular data.frames that contain time series / time like data.
These would detect a date/time object, perform something special. The user could also provide the date/time variable of interest.
Some example mts objects include fpp2::arrivals
We could also consider the following additional arguments/features:
- n_period (how many periods in your time series do you want?)
- size_period (how long are the periods in your time series.)
This could also be a vector:
- size_period = c(3,10,100) OR
- size_period = c(100) OR
- size_period = c(100,50)
Regarding the use of miss_var_run, it might also be useful to summarise the frequency of the size of the runs.
So, ideally, I want variables like:
|| run_size | n_missing | n_complete | prop_missing | prop_complete ||
Possible implementation for multiple spans
pedestrian %>%
mutate(weekday = if_else(Day == "Saturday" | Day == "Sunday",
true = "weekend",
false = "weekday")) %>%
group_by(Sensor_Name,
weekday) %>%
miss_ts_summary(time_index = Date_Time,
variable = Hourly_Counts
naniar:::add_period_counter(period_length = ) %>%
dplyr::group_by(period_counter) %>%
dplyr::tally(is.na(Hourly_Counts))
dplyr::rename(n_miss = n) %>%
dplyr::mutate(n_complete = length(dat_ts) - n_miss,
prop_miss = n_miss / period,
prop_complete = 1 - prop_miss)