Shiny app grid

[wfmackey]

Settings and policy grid below. Ideally we would run each 10 times, so ~70,000 runs.

library(covid19.model.sa2)
library(hutils)
library(data.table)
library(magrittr)

# Default policy settings:

default_policies <- set_policypars(
  # shops
  supermarkets_open = TRUE,
  max_persons_per_supermarket = 50,
  # cafes
  cafes_open = TRUE, 
  # schools
  schools_open = TRUE,
  only_Year12 = FALSE,
  school_days_per_wk = 5,
  # workplaces
  workplaces_open = .70,
  workplace_size_max = 30,
  # contact tracing
  do_contact_tracing = TRUE,
  contact_tracing_days_before_test = 1,
  contact_tracing_days_until_result = 2,
  contact_tracing_success = 0.7,
  contact_tracing_only_sympto = FALSE,
  # major events
  max_persons_per_event = 1,   # not used
  n_major_events_weekday = 0,  # not used
  n_major_events_weekend = 0,  # not used
  # misc
  travel_outside_sa2 = TRUE
)

# Default epis
epis <- set_epipars(
  incubation_distribution = "pois",
  incubation_mean = 8, # note this this is the communicable period
  illness_distribution = "pois",
  illness_mean = 15,
  q_household = 0.20, 
  a_household_rate = 1, 
  q_workplace = 0.10, 
  a_workplace_rate = 0.8,
  q_school = 0.0003, 
  a_schools_rate = 1, 
  q_supermarket = .001,
  q_places = 0.01
)

# Policy grid:
PolicyGrid <-
  CJ(schools_open = c(FALSE, TRUE),
     contact_tracing_days_before_test = c(0:1),
     contact_tracing_days_until_result = c(3L, 5L),
     cafes_open = c(FALSE, TRUE),
     q_places = c(0.05, 0.1, 0.2),
     supermarkets_open = TRUE,
     max_persons_per_supermarket = c(5, 30),
     q_supermarket = c(0.0025, 0.01, 0.05),
     workplaces_open = c(0, 0.5, 0.75, 1),
     workplace_size_max = c(50L, 500L),
     q_workplace = c(.01, .1, .2))

[HughParsonage]

Can you confirm the dimensions of the Grid?

[HughParsonage]

Instead of each run 10 times, would it be preferable to modify the incubation parameters (mean/distribution)?

[wfmackey]

The issue with that adjustment is it’s (expected) distortion of the initial number of active cases. Unless that’s changed somehow, it’s better to show runs using the same epi settings

[HughParsonage]

No problem I will do that -- though bear in mind that that issue is most salient in the midst of the peak. When the number of new cases hasn't changed much it becomes less of an issue.

[wfmackey]

True. We should run on June 1 numbers and on today’s numbers. (Ideally we could run on a flat “500 active cases per state” scenario — is that possible?)

[HughParsonage]

Yep it's possible.