Add duke forest homes data (#53)

Author: mine-cetinkaya-rundel

NEWS.md

@@ -2,7 +2,7 @@
 
 * Added new datasets:
   * `lizard_run` and `lizard_habitat` (thanks Stephen Adolph!)
-  * `daycare_fines`, `biontech_adolescents`, `nyc_marathon`, `epa2021`
+  * `daycare_fines`, `biontech_adolescents`, `nyc_marathon`, `epa2021`, `duke_forest`
 * Add tests for some of the new datasets
 * Better terminology: `sex_discrimination`
 

R/data-duke_forest.R

@@ -0,0 +1,47 @@
+#'  Houses for sale in Duke Forest, Durham, NC in Nov 2020
+#'
+#'  Data on houses that were for sale in the Duke Forest neighborhood of
+#'  Durham, NC in November 2020.
+#'
+#' @format A data frame with 98 rows and 13 variables.
+#' \describe{
+#'    \item{address}{Address of house.}
+#'    \item{price}{Listing price, in USD.}
+#'    \item{bed}{Number of bedrooms.}
+#'    \item{bath}{Number of bathrooms.}
+#'    \item{area}{Area of home, in square feet.}
+#'    \item{type}{Type of home (all are Single Family).}
+#'    \item{year_built}{Year the home was built.}
+#'    \item{heating}{Heating sytem.}
+#'    \item{cooling}{Cooling system (`central` or `other`).}
+#'    \item{parking}{Type of parking available and number of parking spaces.}
+#'    \item{lot}{Area of the entire property, in acres.}
+#'    \item{hoa}{If the home belongs to an Home Owners Association, the associted fee (`NA` otherwise).}
+#'    \item{url}{URL of the listing.}
+#' }
+#' @examples
+#'
+#' library(ggplot2)
+#'
+#' # Number of bedrooms and price
+#' ggplot(duke_forest, aes(x = as.factor(bed), y = price)) +
+#'   geom_boxplot() +
+#'   labs(
+#'     x = "Number of bedrooms",
+#'     y = "Listing price (USD)",
+#'     title = "Homes for sale in Duke Forest, Durham, NC",
+#'     subtitle = "Data are from November 2020"
+#'     )
+#'
+#' # Area and price
+#' ggplot(duke_forest, aes(x = area, y = price)) +
+#'   geom_point() +
+#'   labs(
+#'     x = "Area (square feet)",
+#'     y = "Listing price (USD)",
+#'     title = "Homes for sale in Duke Forest, Durham, NC",
+#'     subtitle = "Data are from November 2020"
+#'     )
+#'
+#' @source Data were collected from [Zillow.com](https://www.zillow.com/) in November 2020.
+"duke_forest"

data-raw/duke_forest/duke-forest.csv

@@ -0,0 +1,25 @@
+# load packages ----------------------------------------------------------------
+
+library(tidyverse)
+library(lubridate)
+
+# load data -------------------------------------------------------------------
+
+duke_forest_raw <- read_csv("data-raw/duke_forest/duke-forest.csv")
+
+# clean data -------------------------------------------------------------------
+
+duke_forest <- duke_forest_raw %>%
+  mutate(
+    cooling = case_when(
+      cooling == "Central" ~ "central",
+      cooling == "Central, Other" ~ "central",
+      TRUE ~ "other"
+      )
+    ) %>%
+  rename(url = link) %>%
+  relocate(url, .after = last_col())
+
+# Save ------------------------------------------------------------------------
+
+usethis::use_data(duke_forest, overwrite = TRUE)

data-raw/duke_forest/duke_forest-dataprep.R

@@ -87,6 +87,7 @@ reference:
   - dream
   - drone_blades
   - drug_use
+  - duke_forest
   - earthquakes
   - ebola_survey
   - elmhurst