In wimmyteam/conisi: What the Package Does (One Line, Title Case)

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Installation

You can install the released version of conisi from github with:

devtools::install_github(
  "wimmyteam/conisi",
  ref="master"
  )

Run a single simulation

Load the parameters

  file_path <- system.file("test-data", "manually_edited_parameters.csv", package="conisi")
  # You would normally provide your own parameter file

  parameters <- conisi::load_parameters(
    file_path, 
    sim_name = "sa_37", 
    start_date = lubridate::ymd("2020-03-10")
  )

Set the size of simulated population and length of simulation

  pop <- 1000000 #The size of the population to simulate
  days <- 100 #The number of days to simulate

Run the simulation and calculate pandemic metrics

  mod_result <- COVIDmodel_run_and_mutate(par_table, pop, days) # Runs the simulation

Calculate the RMSE

  # Load the target data
  load(system.file("test-data", "observed_data.RData", package="conisi")) #Loads test_data

  # start_date determines which date t = 0 from the model output will be matched to.
  start_date <- test_data$date[1]

  rmse <- modelrmse(modelOutput = mod_result,
                    start_date,
                    test_data,
                    weights = c(1, 0.5)
  )

Run many simulations (perhaps using azure)

  # Get the parameters somewhere, needs "experiment" column.
  parameters <- get_par_table()

  # TODO Update the conisi::load_parameters function to allow loading many experiments from a single file.

  # The conisi package uses the foreach package to run simulations in parallel.
  # This required a bit of preparation
  availablecores <- 2
  doParallel::registerDoParallel(cores = availablecores)

  ## Alternatively to run in the cloud on Azure:
  # library(doAzureParallel)
  # setCredentials("../credentials.json")
  # cluster <- makeCluster("cluster.json")
  # registerDoAzureParallel(cluster)
  # getDoParWorkers()

  # Actually run the simulations
  mod_result <- COVIDmodel_run_and_mutate_many(xparm_table, pop, 100)

  ## If using azure, remember to turn off the cluster when you are done
  # stopCluster(cluster)