add tests/ folder with testdata

NIEHS · Jun 5, 2024 · d9a22f2 · d9a22f2
1 parent 4b07723
commit d9a22f2
Showing 1 changed file with 72 additions and 49 deletions.
diff --git a/tests/testdata/testdata_creation.Rmd b/tests/testdata/testdata_creation.Rmd
@@ -8,10 +8,12 @@ output: html_document
 Configuration
 
 ```{r}
-triangle <- rbind(c(-79.19,35.6),
-                  c(-79.19, 36.11),
-                  c(-78.39, 36.11),
-                  c(-78.39, 35.6)) |>
+triangle <- rbind(
+  c(-79.19, 35.6),
+  c(-79.19, 36.11),
+  c(-78.39, 36.11),
+  c(-78.39, 35.6)
+) |>
   terra::vect("polygons", crs = "epsg:4326")
 ```
 
@@ -22,11 +24,13 @@ inv <- "../../input/weatherunderground/Raleigh_Stations_inventory.csv" |>
   read.csv() |>
   sf::st_as_sf(coords = c("lon", "lat"), crs = 4326, remove = FALSE)
 inv$ts_utc <- as.POSIXct(inv$ts_utc,
-                         format = "%Y-%m-%d %H:%M:%S",
-                         tz = "UTC")
+  format = "%Y-%m-%d %H:%M:%S",
+  tz = "UTC"
+)
 inv$te_utc <- as.POSIXct(inv$te_utc,
-                         format = "%Y-%m-%d %H:%M:%S",
-                         tz = "UTC")
+  format = "%Y-%m-%d %H:%M:%S",
+  tz = "UTC"
+)
 samp <- sample(seq(1, nrow(inv), 1), size = 150, replace = FALSE)
 inv <- inv[samp, ]
 ```
@@ -46,47 +50,59 @@ Weather Underground testdata
 ```{r}
 setwd("../..")
 wu <- load_wu(config$ts, config$te, config$area, config$wu_inv)
-keep <- c("stationID",
-          "timezone",
-          "obsTimeUtc",
-          "lat",
-          "lon",
-          "humidityAvg",
-          "tempAvg")
+keep <- c(
+  "stationID",
+  "timezone",
+  "obsTimeUtc",
+  "lat",
+  "lon",
+  "humidityAvg",
+  "tempAvg"
+)
 wu <- wu[, ..keep]
 ```
 
 Create simulated dataset from raw wu
 
 ```{r}
 # replace stationID
-id <- data.table(stationID = unique(wu$stationID),
-                 lon = unique(wu$lon),
-                 lat = unique(wu$lat),
-                 new = as.character(1:length(unique(wu$stationID))))
+id <- data.table(
+  stationID = unique(wu$stationID),
+  lon = unique(wu$lon),
+  lat = unique(wu$lat),
+  new = as.character(seq_along(unique(wu$stationID)))
+)
 # add noise to lat and lon
 id$lat <- id$lat + rnorm(nrow(id), 0, 0.005)
 id$lon <- id$lon + rnorm(nrow(id), 0, 0.005)
-wu_sim <- merge(wu[, c("stationID",
-                       "timezone",
-                       "obsTimeUtc",
-                       "humidityAvg",
-                       "tempAvg")],
-                id,
-                by = "stationID")
+wu_sim <- merge(
+  wu[, c(
+    "stationID",
+    "timezone",
+    "obsTimeUtc",
+    "humidityAvg",
+    "tempAvg"
+  )],
+  id,
+  by = "stationID"
+)
 wu_sim$new <- NULL
 lubridate::second(wu_sim$obsTimeUtc) <- sample(0:59,
-                                               nrow(wu_sim),
-                                               replace = TRUE)
+  nrow(wu_sim),
+  replace = TRUE
+)
 lubridate::minute(wu_sim$obsTimeUtc) <- sample(0:59,
-                                               nrow(wu_sim),
-                                               replace = TRUE)
+  nrow(wu_sim),
+  replace = TRUE
+)
 wu_sim$humidityAvg <- wu_sim$humidityAvg + sample(-5:5,
-                                                  nrow(wu_sim),
-                                                  replace = TRUE)
+  nrow(wu_sim),
+  replace = TRUE
+)
 wu_sim$tempAvg <- wu_sim$tempAvg + sample(-5:5,
-                                          nrow(wu_sim),
-                                          replace = TRUE)
+  nrow(wu_sim),
+  replace = TRUE
+)
 saveRDS(wu_sim, "wu_raw_simulated_testdata.rds")
 ```
 
@@ -98,26 +114,34 @@ PurpleAir testdata
 pa_file <- "../../input/rtp/pa_20210720_20210727.csv"
 pa <- load_pa(config$ts, config$te, config$area, storage_file = pa_file)
 head(pa)
-id <- data.table(sensor_index = unique(pa$sensor_index),
-                 longitude = unique(pa$longitude),
-                 latitude = unique(pa$latitude),
-                 new = as.character(1:length(unique(pa$sensor_index))))
+id <- data.table(
+  sensor_index = unique(pa$sensor_index),
+  longitude = unique(pa$longitude),
+  latitude = unique(pa$latitude),
+  new = as.character(seq_along(unique(pa$sensor_index)))
+)
 # add noise to latitude and longitude
 id$latitude <- id$latitude + rnorm(nrow(id), 0, 0.005)
 id$longitude <- id$longitude + rnorm(nrow(id), 0, 0.005)
-pa_sim <- merge(pa[, c("sensor_index",
-                       "time_stamp",
-                       "humidity",
-                       "temperature")],
-                id,
-                by = "sensor_index")
+pa_sim <- merge(
+  pa[, c(
+    "sensor_index",
+    "time_stamp",
+    "humidity",
+    "temperature"
+  )],
+  id,
+  by = "sensor_index"
+)
 pa_sim$new <- NULL
 pa_sim$humidity <- pa_sim$humidity + sample(-5:5,
-                                                  nrow(pa_sim),
-                                                  replace = TRUE)
+  nrow(pa_sim),
+  replace = TRUE
+)
 pa_sim$temperature <- pa_sim$temperature + sample(-5:5,
-                                          nrow(pa_sim),
-                                          replace = TRUE)      
+  nrow(pa_sim),
+  replace = TRUE
+)
 saveRDS(pa_sim, "pa_raw_simulated_testdata.rds")
 ```
 
@@ -128,4 +152,3 @@ GHCNh testdata
 ghcnh <- download_ghcnh(config$ts, config$te, config$area)
 saveRDS(ghcnh, "ghcnh_testdata.rds")
 ```
-