Skip to content

Latest commit

 

History

History
136 lines (100 loc) · 2.84 KB

chapter1.md

File metadata and controls

136 lines (100 loc) · 2.84 KB
title description free_preview
Hot-deck Imputation
Learner will be able to impute values based on how similar those values are to other values using R.
true

Impute by Matching

type: VideoExercise
key: a297bfdc4c
xp: 50

@projector_key 8a75340ffe779b8d3cfdd8597519ffe0

Using HotDeck in R

type: NormalExercise
key: a52f360854
xp: 100

Here is a quick exercise to learn how to use the hotdeck procedure in R. We will use the mammal sleep data that is pre-loaded in the VIM package. A data frame with 62 observations on the following 10 variables:

  • BodyWgt a numeric vector
  • BrainWgt a numeric vector
  • NonD a numeric vector
  • Dream a numeric vector
  • Sleep a numeric vector
  • Span a numeric vector
  • Gest a numeric vector
  • Pred a numeric vector
  • Exp a numeric vector
  • Danger a numeric vector

@instructions

  1. Install the VIM package.
  2. Load the sleep data.
  3. Choose the 'BrainWgt' variable to sort the data before imputation.
  4. Impute the missing variables using hotdeck.

@hint Use the "ord_var" to choose the "BrainWgt" variable.

@pre_exercise_code

install.packages('VIM')
library('VIM')
data(sleep)

@sample_code

install.packages('VIM')
library('VIM')
data(sleep)
sleepI2 <- hotdeck(...)

@solution

install.packages('VIM')
library('VIM')
data(sleep)
sleepI2 <- hotdeck(sleep,ord_var="BodyWgt")

@sct

success_msg('Good job on completing the exercise!', praise = FALSE)

Simple Imputation Approaches in R

type: NormalExercise
key: fd410340f7
xp: 100

Now we will practice the simple imputation techniques that we have learned in R. We will try a few of these techniques out on a dataset about class grades. For this data, the recorded values are the average of sub-components: e.g the Tutorial variable is the average of all tutorials, the Final exam variable is the average of all questions in the final, written exam.

@instructions First let's visualize the missing variables. Create an aggregation plot using VIM. Does there appear to be a lot of missing values or a little?

Next, let's try mean imputation for the missing values. Try using mean imputation with the HMSIC package for the two variables that have missing data.

@hint

@pre_exercise_code

library(VIM)
library(Hmisc)

data <- read.csv(file = "class-grades.csv", header = TRUE) #load the data

@sample_code

agg_data <- (, numbers = TRUE, prop = c(TRUE, FALSE))

data$ImputedFinal <- (, )

data$ImputedTakeHome <- (, )

@solution

#aggregation plots
agg_data <- aggr(data, numbers = TRUE, prop = c(TRUE, FALSE))

#mean imputation
data$ImputedFinal <- impute(data$Final, mean)

data$ImputedTakeHome <- impute(data$TakeHome, mean)

@sct

#User should use impute()

#User should use agg()

print('Great job getting the exercise done!')