epinowcast error when multiple time series are used: Mismatch in dimension declared and found in context; about epinowcast

[jbracher]

Thanks @seabbs, the example worked without errors. Will let you know in case I have more issue. Also thanks for tweaking the code!

[seabbs]

Thanks Johannes, this certainly seems like a bug related to the recent issues with cmdstan initialisation.

I'll investigate the reprex (thank you) and circle back.

[seabbs]

Yup this was another initialisation issue. Thank you for reporting.

I slightly tweaked your code to make it a little simpler (I don't think your reporting model was very sensible in most settings I can imagine as it had a group-specific hazard intercept on top of group-specific log normal delay)).

# Load data.table and ggplot2
library(data.table)

# Use 2 cores
options(mc.cores = 2)

# Load and filter germany hospitalisations
nat_germany_hosp <-
  germany_covid19_hosp[location == "DE"][age_group %in% c("00+", "80+")]
nat_germany_hosp <- enw_filter_report_dates(
  nat_germany_hosp,
  latest_date = "2021-10-01"
)
# Make sure observations are complete
nat_germany_hosp <- enw_complete_dates(
  nat_germany_hosp,
  by = c("location", "age_group")
)
# Make a retrospective dataset
retro_nat_germany <- enw_filter_report_dates(
  nat_germany_hosp,
  remove_days = 40
)
retro_nat_germany <- enw_filter_reference_dates(
  retro_nat_germany,
  include_days = 40
)
# Get latest observations for the same time period
latest_obs <- enw_latest_data(nat_germany_hosp)
latest_obs <- enw_filter_reference_dates(
  latest_obs,
  remove_days = 40, include_days = 20
)
# Preprocess observations (note this maximum delay is likely too short)
pobs <- enw_preprocess_data(retro_nat_germany, by = "age_group", max_delay = 20)

# set reference time indexed reporting process model
reference_module <- enw_reference(
  ~ 1 + (1 | .group), distribution = "lognormal", data = pobs
)

# set reporting time indexed reporting process model
report_module <- enw_report(~ 1, data = pobs)

# set expectation module
expectation_module <- enw_expectation(
  ~ rw(week, by = .group),
  observation = ~ (1 | day_of_week:.group),
  data = pobs
)

# Fit the nowcast model and produce a nowcast
# Note that we have reduced samples for this example to reduce runtimes
nowcast <- epinowcast(pobs,
                      reference = reference_module,
                      report = report_module,
                      expectation = expectation_module,
                      fit = enw_fit_opts(
                        save_warmup = FALSE, pp = TRUE,
                        chains = 2, iter_warmup = 500, iter_sampling = 500,
                        max_treedepth = 12, adapt_delta = 0.95
                      )
)

plot(nowcast, latest_obs) + facet_wrap(vars(age_group), scales = "free")

The code for this is on seabbs/seabbs/issue281 - just assembling some unit tests for this now and then will PR.

FYI if you use multi-threading on this model you should get a linear speed up. Also the performance of daily random walks leaves a lot to be desired. Happy to hear suggests for improvements but ultimately likely needs a lot of profiling to work out how to do better.