Merge pull request #19 from mlr-org/ames

feat: add ames data set

Author: mllg

DESCRIPTION

@@ -28,6 +28,7 @@ Suggests:
     mlr3 (>= 0.13.3)
 Encoding: UTF-8
 LazyData: true
+LazyDataCompression: xz
 NeedsCompilation: no
 Roxygen: list(markdown = TRUE)
-RoxygenNote: 7.2.1
+RoxygenNote: 7.2.3

R/ames_housing.R

@@ -0,0 +1,24 @@
+#' @title House Sales in Ames, Iowa
+#'
+#' @name ames_housing
+#' @aliases mlr_tasks_ames_housing
+#'
+#' @description
+#' Regression task to predict house sale prices for Ames, Iowa.
+#'
+#' Contains 80 features and 2930 observations.
+#' Target column is `"Sale_Price"`.
+#'
+#' @docType data
+#' @keywords data
+#' @examples
+#' data("ames_housing", package = "mlr3data")
+#' str(ames_housing)
+NULL
+
+get_ames_housing_task = function() {
+  b = as_backend("ames_housing")
+  task = mlr3::TaskRegr$new("ames_housing", b, target = "Sale_Price", label = "Ames House Sales")
+  b$hash = task$man = "mlr3data::mlr_tasks_ames_housing"
+  task
+}

R/energy_usage.R

@@ -0,0 +1,16 @@
+#' @title Power Consumption of Kitchen Appliances in Ames, Iowa
+#'
+#' @name energy_usage
+#'
+#' @description
+#' Data for power consumption of kitchen appliances in Ames, Iowa.
+#' Extends the ames data set.
+#'
+#' Contains 720 features and 2930 observations.
+#'
+#' @docType data
+#' @keywords data
+#' @examples
+#' data("energy_usage", package = "mlr3data")
+#' str(energy_usage)
+NULL

R/zzz.R

@@ -4,6 +4,7 @@
 register_tasks = function() {
   x = getFromNamespace("mlr_tasks", ns = "mlr3")
 
+  x$add("ames_housing", get_ames_housing_task)
   x$add("bike_sharing", get_bike_sharing_task)
   x$add("ilpd", get_ilpd_task)
   x$add("kc_housing", get_kc_housing_task)

data-raw/ames_housing.R

@@ -0,0 +1,3 @@
+url = "https://raw.githubusercontent.com/ja-thomas/extend_ames_housing/main/data/ames_dirty.csv"
+ames_housing = data.table::fread(url, stringsAsFactors = TRUE)
+usethis::use_data(ames_housing, overwrite = TRUE, compress = "xz")

data-raw/bike_sharing.R

@@ -5,13 +5,13 @@ download.file("https://archive.ics.uci.edu/ml/machine-learning-databases/00275/B
 bike_sharing = setDT(read.csv(unz(tmp, filename = "hour.csv"), as.is = TRUE))
 bike_sharing = setnames(bike_sharing, c("instant", "date", "season", "year", "month", "hour", "holiday", "weekday",
   "working_day", "weather", "temperature", "apparent_temperature", "humidity", "windspeed", "casual", "registered",
-  "count")) 
+  "count"))
 
 bike_sharing[, holiday := as.logical(holiday)]
 bike_sharing[, working_day := as.logical(working_day)]
 bike_sharing[, season := factor(season, labels = c("winter", "spring", "summer", "fall"))]
-bike_sharing[, weather :=  as.factor(bike_sharing$weather)] 
+bike_sharing[, weather :=  as.factor(bike_sharing$weather)]
 
 bike_sharing = bike_sharing[, .SD, .SDcols = !c("instant", "casual", "registered")]
 
-usethis::use_data(bike_sharing, overwrite = TRUE)
+usethis::use_data(bike_sharing, overwrite = TRUE, compress = "xz")

data-raw/energy_usage.R

@@ -0,0 +1,3 @@
+url = "https://raw.githubusercontent.com/ja-thomas/extend_ames_housing/main/data/energy_usage.csv"
+energy_usage = data.table::fread(url, stringsAsFactors = TRUE)
+usethis::use_data(energy_usage, overwrite = TRUE, compress = "xz")

data-raw/ilpd.R

@@ -9,4 +9,4 @@ levels(ilpd$diseased) = c("yes", "no")
 
 data.table::setDF(ilpd)
 
-usethis::use_data(ilpd, overwrite = TRUE)
+usethis::use_data(ilpd, overwrite = TRUE, compress = "xz")

data-raw/kc_housing.R

@@ -10,6 +10,6 @@ kc_housing[, date := anytime::anytime(date)]
 kc_housing[yr_renovated == 0, yr_renovated := NA_integer_]
 kc_housing[sqft_basement == 0, sqft_basement := NA_integer_]
 kc_housing[, waterfront := as.logical(waterfront)]
-setDF(kc_housing)
+data.table::setDF(kc_housing)
 
-usethis::use_data(kc_housing, overwrite = TRUE)
+usethis::use_data(kc_housing, overwrite = TRUE, compress = "xz")

data-raw/moneyball.R

@@ -4,8 +4,8 @@ odata = OMLData$new(41021)
 moneyball = odata$data
 to_integer = c("Year", "RS", "RA", "W")
 moneyball[, (to_integer) := lapply(.SD, as.integer), .SDcols = to_integer]
-setnames(moneyball, new = tolower(colnames(moneyball)))
+data.table::setnames(moneyball, new = tolower(colnames(moneyball)))
 
-setDF(moneyball)
+data.table::setDF(moneyball)
 
-usethis::use_data(moneyball, overwrite = TRUE)
+usethis::use_data(moneyball, overwrite = TRUE, compress = "xz")