replace dist_spec() code with get_parameter() + Gamma() or LogNormal()

[avallecam]

replace

tutorials-early/episodes/read-delays.Rmd

Lines 814 to 822 in 8dcbabf

	serial_interval_covid <-
	dist_spec(
	mean = covid_serialint$summary_stats$mean,
	sd = covid_serialint$summary_stats$sd,
	max = covid_serialint_discrete_max,
	distribution = "lognormal"
	)
	serial_interval_covid

from

# update: EpiNow2 1.4.9 ---------------------------------------------------

si_x <- seq(1L, to = covid_serialint_discrete_max, by = 1L)

serial_interval_covid_update <- 
  EpiNow2::dist_spec(pmf = covid_serialint_discrete$prob_dist$d(si_x))

serial_interval_covid_update
#> - nonparametric distribution
#>   PMF: [0.0073 0.1 0.2 0.19 0.15 0.11 0.075 0.051 0.034 0.023 0.016 0.011 0.0076 0.0053 0.0037 0.0027 0.0019 0.0014 0.001 0.00074 0.00055 0.00041 0.00031]

key learning goals

• a simple way to plug in epiparameter is using the pmf
• we can also use functions like LogNormal() or Gamma() but these require more (unnecessary) parameters
• we need max parameter to plug in LogNormal() output to epinow()
• the max parameter can come from a discretized continuous function as an epidist object

[avallecam]

change my mind. the simpler input is still using the explicit distribution parameters

# new branch: is this a simpler and valid alternative? --------------------------------

covid_serialint_parameters <- epiparameter::get_parameters(covid_serialint)

covid_serialint_parameters
#>   meanlog     sdlog 
#> 1.3862617 0.5679803

serial_interval_covid_branch_lognormalmax <- 
  EpiNow2::LogNormal(
    meanlog = covid_serialint_parameters["meanlog"],
    sdlog = covid_serialint_parameters["sdlog"],
    max = covid_serialint_discrete_max)

serial_interval_covid_branch_lognormalmax
#> - lognormal distribution (max: 23):
#>   meanlog:
#>     1.4
#>   sdlog:
#>     0.57

updated

key learning goals

• LogNormal() or Gamma() can be used for fixed and uncertain distributions
• LogNormal() or Gamma() require lesser steps than pmf
• LogNormal() or Gamma() must have the max parameter to plug in their output to epinow()
• the max parameter can come from a discretized continuous function as an epidist object

comparison of steps

epidist

• get_parameters()
• discrete
  • max

compared to

epidist

• discrete
• max
• seq
• density
• pmf

[avallecam]

this update depends on the merging of epiforecasts/EpiNow2#504

[avallecam]

fixed in #80