Merge pull request #86 from DylanCarbone/pull_requests

Pull requests

NAMESPACE

@@ -5,7 +5,6 @@ export(CompositeTrend)
 export(GAM_smoothing)
 export(bma)
 export(bootstrap_indicator)
-export(detect_jags)
 export(get_bmaBUGScode)
 export(lambda_indicator)
 export(msi)

R/bayesian_meta_analysis.R

@@ -45,7 +45,7 @@
 #' @examples 
 #' 
 #' # Only run if there is a JAGS installation
-#' if(detect_jags()){
+#' if(suppressWarnings(runjags::testjags(silent = TRUE)$JAGS.found)){
 #' 
 #' # Create some example data in the format required
 #' data <- data.frame(species = rep(letters, each = 50),
@@ -91,12 +91,6 @@ stop("No installation of JAGS has been detected. You can install JAGS from https
       call. = FALSE)
 }
 
-# Check if jagsUI is installed
-if (!requireNamespace("jagsUI", quietly = TRUE)) {
-stop("Package 'jagsUI' is needed for the 'bma' function to work. Please install this from CRAN.",
-      call. = FALSE)
-}
-
 # Check to see that column names are present
 if (!all(c("species", "year", "index") %in% colnames(data))){
 

R/detect_jags.r

@@ -1,13 +1,6 @@
-#' Detect JAGS installation
-#' 
-#' @description small function that detects whether JAGS is installed.
-#' @return TRUE or FALSE, indicating whether the JAGS installation has been detected 
+# Internal function to detect JAGS installation.
 #' @importFrom runjags testjags
-#' @export
-#' @examples
 #' 
-#' detect_jags()
-
 detect_jags <- function(){
     return(suppressWarnings(runjags::testjags(silent = TRUE)$JAGS.found))
 }

R/rescale_species.R

@@ -16,7 +16,38 @@
 #' @return A matrix. In each row we have the year, the species and the
 #'         scaled value. There is also an additional column, 'geomean' giving
 #'         the geometric mean for each year.
+#' 
 #' @export
+#' 
+#' @examples
+#' 
+#' # We will create fake data which will mimic a summarised output from a sparta package occupancy detection model. Please see the vignette for more details.
+#' 
+#' # Create a matrix with random data using set.seed for reproducibility
+#' set.seed(123)
+#' data <- matrix(rnorm(50 * 27), nrow = 50, ncol = 27)
+#' 
+#' # Assign column names
+#' col_names <- c("year", letters[1:26])
+#' dimnames(data) <- list(NULL, col_names)
+#' 
+#' # Create the 'trends_summary' object
+#' trends_summary <- data
+#' 
+#' # Re-scale the data so that the value for all species in the first year is the same (all defaults used).
+#' # This function accounts for species that have no data at the beginning of the dataset by entering them at the geometric mean for that year, this stops them dramatically changing the indicator value in the year they join the dataset.
+#' # It also accounts for species that leave the dataset before the end by holding them at their last value.
+#' # Finally it limits the species values that can be given, preventing extremely high or low values biasing the indicator.
+#' rescale_species(Data = trends_summary)
+#' 
+#' # Now, I have 'messed up' the data a bit by including two species with data missing at the beginning and one species with data missing at the end. We also have one species with some very high values.
+#' trends_summary[1:3, 'a'] <- NA
+#' trends_summary[1:5, 'b'] <- NA
+#' trends_summary[2:4, 'c'] <- 1000
+#' trends_summary[45:50, 'd'] <- NA
+#' 
+#' # Let's run this data again through our scaling function (all defaults used)
+#' rescale_species(Data = trends_summary)
 
 rescale_species <-  function(Data, index = 100, max = 10000,
                              min = 1){

tests/testthat/helper-bayesian_meta_analysis.R

@@ -37,12 +37,6 @@ stop("No installation of JAGS has been detected. You can install JAGS from https
       call. = FALSE)
 }
 
-# Check if jagsUI is installed
-if (!requireNamespace("jagsUI", quietly = TRUE)) {
-stop("Package 'jagsUI' is needed for the 'bma' function to work. Please install this from CRAN.",
-      call. = FALSE)
-}
-
 # Check to see that column names are present
 if (!all(c("species", "year", "index") %in% colnames(data))){
 

tests/testthat/test-bayesian_meta_analysis.R

@@ -1,19 +1,11 @@
 test_that("Does the function stop when there is no jags installation?", {
   # Mock `detect_jags` to return FALSE
-  mockery::stub(bma, "detect_jags", FALSE)
-
-  expect_error(bma(bayesian_meta_analysis_fake_data), regexp = "No installation of JAGS has been detected")
-})
-
-# Test that function responds correctly when 'jagsUI' is not available
-test_that("Does the function stop when 'jagsUI' is not available?", {
-  # Mock `detect_jags` to return TRUE
-  mockery::stub(bma, "detect_jags", TRUE)
-
-  # Mock `requireNamespace` to return FALSE
-  mockery::stub(bma, "requireNamespace", FALSE)
-
-  expect_error(bma(bayesian_meta_analysis_fake_data), regexp = "Package 'jagsUI' is needed")
+  with_mocked_bindings(
+    "detect_jags" = function() FALSE,
+    {
+      expect_error(bma(bayesian_meta_analysis_fake_data), regexp = "No installation of JAGS has been detected")
+    }
+  )
 })
 
 test_that("Does it correctly detect missing columns in the input data?", {

vignettes/BRCindicators.Rmd

@@ -19,11 +19,9 @@ knitr::opts_chunk$set(echo = TRUE,
                       tidy.opts = list(width.cutoff = 60),
                       tidy = FALSE)
 ```
+
 ```{r libraries}
 require(snowfall)
-
-
-to produce an error
 ```
 
 # Introduction
@@ -42,7 +40,7 @@ If you already have yearly estimates of abundance or occurrence for your species
 
 ```{r set_up, eval=FALSE}
 library(devtools)
-#install_github('biologicalrecordscentre/sparta')
+install_github('biologicalrecordscentre/sparta')
 ```
 
 Let's assume you have some raw data already, we can under take occupancy modelling like this