add 3 letter country code

DESCRIPTION

@@ -18,7 +18,6 @@ Depends: R (>= 2.10)
 RoxygenNote: 6.1.1
 Suggests: 
     dplyr,
-    knitr,
     gganimate,
     ggplot2,
     ggthemes,

R/tdor.R

@@ -33,8 +33,8 @@
 #'     \item{Age}{Age (character), typically age in years, but may also be
 #'     a range (if reports are inconsistent), or an approximation (e.g.
 #'     "Approx. 30" or "Under 35").}
-#'     \item{Age_min}{Minium age (numeric) if this can be defined from Age.}
-#'     \item{Age_max}{Maxium age (numeric) if this can be defined from Age.}
+#'     \item{Age min}{Minium age (numeric) if this can be defined from Age.}
+#'     \item{Age max}{Maxium age (numeric) if this can be defined from Age.}
 #'     \item{Photo source}{URL where photo was obtained from for
 #'     tdor.translivesmatter.info website.}
 #'     \item{Date}{Date of death.}
@@ -46,6 +46,7 @@
 #'     \item{State/Province}{The state or province. If `NA`, the state/province
 #'     is part of `Location`.}
 #'     \item{Country}{Country where death occurred.}
+#'     \item{Country code}{ISO-3 country code (three letter code).}
 #'     \item{Latitude}{Latitude of location.}
 #'     \item{Longitude}{Longitude of location.}
 #'     \item{Cause of death}{Category of cause of death.}

data-raw/tdor.R

@@ -1,10 +1,12 @@
+library("countrycode")
 library("dplyr")
 library("janitor")
 library("lubridate")
 library("purrr")
 library("readr")
 library("tidyr")
 library("usethis")
+library("WDI")
 
 # Get Data ----
 # Data sourced from https://bitbucket.org/annajayne/tdor_data/downloads/, unzipped and the contents of the data folder placed in data-raw
@@ -68,12 +70,12 @@ tdor <- files %>%
 
 tdor <- tdor %>%
   # sometimes age is a range; separate into min and max
-  separate(Age, into = c("Age_min", "Age_max"), sep = "-", remove = FALSE) %>%
-  mutate(Age_min = suppressWarnings(ifelse(!is.na(as.numeric(Age_min)),
-                                           as.numeric(Age_min),
+  separate(Age, into = c("Age min", "Age max"), sep = "-", remove = FALSE) %>%
+  mutate(`Age min` = suppressWarnings(ifelse(!is.na(as.numeric(`Age min`)),
+                                           as.numeric(`Age min`),
                                            as.numeric(sub("(Approx. )",
                                                           "", Age)))),
-         Age_max = suppressWarnings(ifelse(!is.na(Age_max), as.numeric(Age_max),
+         `Age max` = suppressWarnings(ifelse(!is.na(`Age max`), as.numeric(`Age max`),
                                            as.numeric(sub("(Approx. |Under )",
                                                           "", Age))))) %>%
   # simple conversion to TDOR period no longer works
@@ -97,7 +99,14 @@ tdor <- tdor %>%
              # 1st October to 30th September from now on
              TRUE ~ ifelse(Month %in% 1:9, Year, Year + 1)
            )) %>%
-  select(Name:Date, Month, Year, TDoR, everything(), -Photo)
+  # country code
+  mutate(`Country code` = countrycode(Country, "country.name", "iso3c")) %>%
+  # logical order
+  select(Name, Age, `Age min`, `Age max`,
+         Date, Month, Year, TDoR,
+         Location, `State/Province`, Country, `Country code`, Latitude, Longitude,
+         `Cause of death`, Description,
+         `Photo source`, `Source ref`, Tweet, Permalink)
 
 use_data(tdor, overwrite = TRUE)
 

vignettes/exploring_2019_data.Rmd

@@ -22,9 +22,9 @@ Load required packages.
 library(dplyr)
 library(ggplot2)
 library(ggthemes)
+library(lubridate)
 library(maps)
 library(tdor)
-library(lubridate)
 ```
 
 ## Some code and graphs to explore the 2019 data
@@ -58,7 +58,7 @@ Where we have ages... what were those ages?
 ```{r deaths_by_age, warning=FALSE}
 
 tdor_subset %>%
-  ggplot(aes(x = (Age_min + Age_max)/2)) + 
+  ggplot(aes(x = (`Age min` + `Age max`)/2)) + 
   geom_bar() +
   ggtitle("Deaths by age") +
   labs(y = "Deaths", x = "Age at death",

vignettes/exploring_data_set.Rmd

@@ -42,15 +42,15 @@ In some cases there is ambiguity over the age, so we have a minimum and a maximu
 
 ```{r deaths_by_age, warning=FALSE}
 tdor %>% 
-  filter(Age_min > 0 & Age_max > 0) %>%
-  ggplot(aes(x = (Age_min + Age_max)/2)) + 
+  filter(`Age min` > 0 & `Age max` > 0) %>%
+  ggplot(aes(x = (`Age min` + `Age max`)/2)) + 
   geom_bar() +
   ggtitle("Deaths by age") +
   labs(y = "Deaths")
 
 tdor %>% 
-  filter(Age_min > 0 & Age_max > 0) %>%
-  ggplot(aes(x = (Age_min + Age_max)/2)) +
+  filter(`Age min` > 0 & `Age max` > 0) %>%
+  ggplot(aes(x = (`Age min` + `Age max`)/2)) +
   geom_histogram(binwidth = 5) +
   ggtitle("Deaths by age") +
   labs(y = "Deaths")