Test the sensitivity of a Libbi model using PRCC

Test the sensitivity of a Libbi model using PRCC

test_sensitivity(model = NULL, obs = NULL, target_time = NULL)

Arguments

model	A LiBBi model object
obs	A character string containing the observation to use to calculate the PRCC.
target_time	Numeric, the time at which to estimate the model sensitivity. If not specified then this will default to the last fitted point that the model has produced output for.

Value

A data frame containing the names of the parameters in the model, the correlation with the outcome and the p value of this correlation

Examples

test_sensitivity
#> function(model = NULL, obs = NULL, target_time = NULL) {
#>   
#>   model_params <- rbi::bi_read(model, type = "param") %>% 
#>     bind_rows(.id = "Parameter")
#>   
#>   model_obs <- rbi::bi_read(model) %>% 
#>     {.[names(.) %in% obs]} %>%  
#>     bind_rows(.id = "Observation") 
#>   if (is.null(target_time)) {
#>     target_time <- model_obs$time %>% 
#>       max
#>   }
#>   
#>   obs <- model_obs %>% 
#>     dplyr::filter(time == target_time)
#> 
#>   
#>   
#>   ## Spread parameters into the form the prcc funciton expects
#>   params <- model_params %>% 
#>     select(Parameter, value, np) %>% 
#>     spread(key = "Parameter", value = "value") %>% 
#>     select(-np) %>% 
#>     select_if(~ var(.) > 0)
#>   
#>   ## Join obs to parameters
#>   sample <- params %>% 
#>     bind_cols(obs %>% 
#>                 select(value) %>% 
#>                 setNames("Observation"))
#>   
#>   prcc <- epi.prcc(sample) %>% 
#>     mutate(Parameter = colnames(params)) %>% 
#>     select(Parameter, gamma, p.value) %>% 
#>     arrange(desc(abs(gamma)))
#>   
#>   
#>   return(prcc)
#> }
#> <environment: namespace:ModelTBBCGEngland>