Solution for the Preparation phase

DESCRIPTION

@@ -4,6 +4,7 @@ Title: Data Cleaning - what else?
 Version: 0.1.0
 Author: Quantargo Instructor
 Maintainer: Quantargo Instructor <courses@quantargo.com>
+Imports: stats
 Description: Some lengthy description which somebody should write
     Sometime.
 License: GPL-3

NAMESPACE

@@ -1 +1,2 @@
 exportPattern("^[[:alpha:]]+")
+importFrom("stats", "quantile")

R/meanimpute.R

@@ -1,5 +1,14 @@
 #' Meanimputation
+#'
+#' Replaces NA values with the mean of \code{x}
+#' 
+#' @param x A numeric vector containing real numbers
+#' @return Imputed vector \code{x} where NAs are replaced with the mean of \code{x}
+#' @examples
+#' meanimpute(x = c(1, NA, 3, 4, 5, 6, 7, 8, NA, 10))
+#' meanimpute(c(1, NA, 3, 4, 5, 6, 7, 8, NA, 10))
 #' @export
+
 meanimpute <- function(x) {
   x[is.na(x)] <- mean(x, na.rm = TRUE)
   x

R/transform_log.R

@@ -0,0 +1,26 @@
+#' Transform_log
+#'
+#' Transforms values to natural logarithm.
+#' 
+#' @param x A numeric vector containing positive real numbers
+#' @return Natural logarithm of the values in \code{x}.
+#' @examples
+#' transform_log(x = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10))
+#' transform_log(c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10))
+#' @export
+
+transform_log <- function(x){
+
+  if( sum(is.na(x)) ){
+    stop("Input x contains value NA")
+  }
+  else if( length(x) == 0 ){
+    stop("Input x is NULL")
+  }  
+  else if( sum(x <= 0)  ){
+    stop("Input x contains non-positive values")
+  }  
+
+  log(x)
+
+}

R/windsorize.R

@@ -1,10 +1,29 @@
 #' Windsorize
 #'
 #' Do some windsorization.
+#' 
+#' @param x A numeric vector containing real numbers
+#' @param p A number between 0 and 1, representing the mass of probability. Values are winsorized outside \code{p}. Default value is 0.9. 
+#' @return Windsorized vector of \code{x}.
+#' @examples
+#' windsorize(x = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10))
+#' windsorize(x = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10), p = 0.8)
+#' windsorize(c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10), 0.8)
 #' @export
+
 windsorize <- function(x, p = .90) {
-  q <- quantile(x, p)
-  x[x >= q] <- q
+
+  if( sum(is.na(x))){
+    stop("Input x contains value NA")
+  }
+  else if( length(x) == 0 ){
+    stop("Input x is NULL")
+  }
+
+  q_lower <- quantile(x, (1-p)/2)
+  q_upper <- quantile(x, p + (1-p)/2)
+  x[x <= q_lower] <- q_lower
+  x[x >= q_upper] <- q_upper
   x
 }
 

tests/testthat.R

@@ -0,0 +1,4 @@
+library(testthat)
+library(datacleaner)
+
+test_check("datacleaner")

tests/testthat/testthat.R

@@ -0,0 +1,19 @@
+library(datacleaner)
+
+test_that("windsorize the input vector", {
+  expect_error(windsorize(c(NULL)), "Input x is NULL")
+  expect_error(windsorize(c(1, 2, NA)), "Input x contains value NA")
+  expect_equal(windsorize(c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10), 0.8), c(1.9, 2, 3, 4, 5, 6, 7, 8, 9, 9.1))
+})
+
+test_that("Takes natural logarithm", {
+  expect_error(transform_log(c(NULL)), "Input x is NULL")
+  expect_error(transform_log(c(1, 2, NA)), "Input x contains value NA")
+  expect_error(transform_log(c(-1, 2, 3)), "Input x contains non-positive values")
+  expect_equal(transform_log(c(1, exp(1))), c(0, 1))
+})
+
+test_that("Takes natural logarithm", {
+  expect_equal(meanimpute(c(1, NA, 3)), c(1, 2, 3))
+  expect_equal(meanimpute(c(1, 2 , 3)), c(1, 2, 3))
+})