make forecasts from sim

make forecasts from sim

# S3 method for fit_pansim
predict(
  object,
  end_date = NULL,
  stoch = NULL,
  stoch_start = NULL,
  keep_vars = c("H", "ICU", "death", "hosp", "incidence", "report", "cumRep",
    "newTests/1000"),
  ensemble = FALSE,
  new_params = NULL,
  Sigma = NULL,
  scale_Sigma = 1,
  ...
)

# S3 method for fit_pansim
plot(
  x,
  predict_args = NULL,
  data = NULL,
  break_dates = NULL,
  dlspace = 1,
  limspace = 10,
  add_tests = FALSE,
  add_ICU_cap = FALSE,
  mult_var = NULL,
  directlabels = TRUE,
  log = TRUE,
  log_lwr = 1,
  ...
)

Arguments

object

a fitted object

end_date

ending date for sim

stoch

stochasticity

stoch_start

stoch starting date

keep_vars

...

ensemble

run ensemble?

new_params

parameters to update in base parameters (e.g. adding stochastic parameters)

Sigma

covariance matrix

scale_Sigma

inflate/deflate covariance matrix

...

extra args (passed to forecast_ensemble)

x

a calibrated object (result from calibrate) or a prediction (from predict.fit_pansim)

predict_args

additional arguments to pass to predict

data

original time series data

break_dates

breakpoints

dlspace

spacing for direct labels (not working)

limspace

extra space (in days) to add to make room for direct labels

add_tests

plot newTests/1000?

add_ICU_cap

include horizontal lines showing ICU capacity?

mult_var

variable in data set indicating multiple forecast types to compare

directlabels

use direct labels?

log

use a log10 scale for the y axis?

log_lwr

lower limit when using log scale

Examples

pp1 <- predict(ont_cal1, keep_vars="Rt")
#> Warning: specifying params_timevar with Relative_value is deprecated: auto-converting (reported once per session)
## example of hacking params
ont_cal2 <- ont_cal1
ont_cal2$forecast_args$base_params["zeta"] <- 4
pp2 <- predict(ont_cal2, keep_vars="Rt")
#> Warning: specifying params_timevar with Relative_value is deprecated: auto-converting (reported once per session)
## if zeta is fitted probably need to hack x$mle2@coef, e.g.
ont_cal3 <- ont_cal1
## increase beta0 (from -0.34) rather than
## mess with zeta, since phenom het isn't
## estimated in this fit
ont_cal3$mle2@fullcoef["params.log_beta0"] <- 0
pp3 <- predict(ont_cal3, keep_vars="Rt")
#> Warning: specifying params_timevar with Relative_value is deprecated: auto-converting (reported once per session)
pp <- dplyr::bind_rows(base=pp1,zeta=pp2,beta0=pp3, .id="fit")
if (require("ggplot2")) {
   ggplot(pp,aes(date,value,colour=fit))+geom_line()
}
#> Loading required package: ggplot2
#> 
#> Attaching package: ‘ggplot2’
#> The following object is masked from ‘package:McMasterPandemic’:
#> 
#>     %+%

if (FALSE) {
## non-pos-def vcov ... ???
predict(ont_cal_2brks,ensemble=TRUE)
}