adding basic infrastructure including testing (currently not passing)

.Rbuildignore

@@ -4,3 +4,4 @@
 ^README\.Rmd$
 ^cran-comments\.md$
 ^\.github$
+^data-raw$

.gitignore

@@ -2,3 +2,4 @@
 .Rhistory
 .RData
 .Ruserdata
+^data-raw$

DESCRIPTION

@@ -20,4 +20,8 @@ LazyData: true
 Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.1.1
 Imports: 
-    wnvdata
+    wnvdata,
+    dplyr,
+    readr
+Depends: 
+    R (>= 2.10)

NAMESPACE

@@ -1,2 +1,3 @@
 # Generated by roxygen2: do not edit by hand
 
+export(fit_models)

R/data.R

@@ -0,0 +1,8 @@
+#' sampledat
+#'
+#' simulated data on annual numbers of human cases of neuro-invasive and non-neuro-invasive
+#' West Nile Virus in Nebraska counties. It is predictions of a model that was trained on
+#' actual numbers of cases as recorded in CDC's Arbonet database.
+#' @docType data
+#'
+"sampledat"

R/fit_models.R

@@ -0,0 +1,33 @@
+#' Fit all the models and return a list object suitable for DFMIP.
+#'
+#' @param x tibble or data frame that defines the desired prediction targets. Must have at least a FIPS code and year.
+#' @param path_2_case_data character file path to the case data
+#' @param case_type character type of case data
+#' @param case_variable character name of the case field in the case data
+#'
+#' @details This function uses the rows in x to define the prediction
+#' targets for the models. There must be at least two fields, the 5 digit fips code
+#' to identify the county as a character variable and a year numeric variable to
+#' identify the year.
+#'
+#' The WNV case data cannot be posted publicly as a part of this package. This package
+#' assumes the case data are available in a subdirectory of the working directory
+#' called `data-raw`, which has rows
+#' identified by the same fips codes and a year variable. Case counts are assumed
+#' to be in a variable called "cases". There must be at least one row for each
+#' fips code in `x`.
+#'
+#'
+#' @return a list of containing predicted cases for the target in the format expected by `DFMIP`
+#' @export
+fit_models <- function(x, path_2_case_data = here::here("data-raw/wnv_by_county.csv"),
+                       case_type = c("neuro", "all"),
+                       case_variable = "cases"){
+  if(is.null(x) | !is.data.frame(x) | !tibble::is.tibble(x)){
+    stop("x must be a data frame or tibble.")
+  }
+  if(!file.exists(path_2_case_data)){
+    stop("Please provide a valid path to the WNV case data.")
+  }
+  return(list())
+}

data-raw/sampledat.R

@@ -0,0 +1,15 @@
+## code to prepare `sampledat` dataset goes here
+# make sampledat from fitted model
+# assumes an input file with the fitted values in the cases column.
+#sampledat <- readr::read_csv("data-raw/sampledat.csv")
+sampledat <- readr::read_csv("data-raw/predictionsthrough2018.csv") %>%
+  rename(cases = pred)
+# fix arthur county with bayesian posterior for 16 samples of 0 from a poisson distribution
+# given a 1, 1 gamma prior we have gamma(1, 17) as the posterior. This
+# still has a probability 0.52 of observing 16 zeros at the median.
+# So the median is 0.04077336
+sampledat <- dplyr::bind_rows(sampledat, dplyr::tibble(year = 2002:2019,
+                             County = "Arthur",
+                             cases = 0.0477336))
+usethis::use_data(sampledat, overwrite = TRUE)
+

tests/testthat/test-fit_models.R

@@ -0,0 +1,9 @@
+test_that("data checks work", {
+  expect_error(fit_models(x = NULL), "x must be a data frame or tibble.")
+  expect_error(fit_models(x = sampledat, path_2_case_data = "~"),
+               "Please provide a valid path to the WNV case data.")
+})
+
+test_that("return value valid", {
+  expect_type(fit_models(x = sampledat), "list")
+})