preprocessing.md

Preprocessing

Setup

library(caret)

## Loading required package: lattice

## Loading required package: ggplot2

library(kernlab)

## 
## Attaching package: 'kernlab'

## The following object is masked from 'package:ggplot2':
## 
##     alpha

library(RANN)
data(spam)

Split the Data

inTrain <- createDataPartition(y = spam$type, p = 0.75, list = FALSE)
training <- spam[inTrain,]
testing <- spam[-inTrain,]

hist(training$capitalAve, main = "", xlab= "Ave. Capital run length")

Spread variables across all length

mean(training$capitalAve)

## [1] 5.374265

sd(training$capitalAve)

## [1] 35.12402

Very high standard deviation

Standardizing

trainCapAve <- training$capitalAve
trainCapAveS <- (trainCapAve - mean(trainCapAve))/sd(trainCapAve)
mean(trainCapAveS)

## [1] -8.079203e-18

Standardize test set

When standardizing a test set we need to use the train mean and train std We cannot use test mean and test std!

testCapAve <- testing$capitalAve
testCapAveS <- (testCapAve - mean(trainCapAve)) / sd(trainCapAve)
mean(testCapAveS)

## [1] -0.0208165

sd(testCapAveS)

## [1] 0.5139097

Standardization using preProcess function

preObj <- preProcess(training[,-58], method = c("center", "scale"))
trainCapAveS <- predict(preObj, training[, -58])$capitalAve
mean(trainCapAveS)

## [1] -8.079203e-18

sd(trainCapAveS)

## [1] 1

testCapAveS <- predict(preObj, testing[, -58])$capitalAve
mean(testCapAveS)

## [1] -0.0208165

sd(testCapAveS)

## [1] 0.5139097

Standardizing - preProcess argument

set.seed(323243)
model <- train(type ~ ., data = training, preProcess = c("center", "scale"), method = "glm")
model

## Generalized Linear Model 
## 
## 3451 samples
##   57 predictor
##    2 classes: 'nonspam', 'spam' 
## 
## Pre-processing: centered (57), scaled (57) 
## Resampling: Bootstrapped (25 reps) 
## Summary of sample sizes: 3451, 3451, 3451, 3451, 3451, 3451, ... 
## Resampling results:
## 
##   Accuracy  Kappa    
##   0.909105  0.8106326

Standardizing Box-Cox Transforms

preObj <- preProcess(training[, -58], method = c("BoxCox"))
trainCapAveS <- predict(preObj, training[, -58])$capitalAve
par(mfrow = c(1, 2))
hist(trainCapAveS)
qqnorm(trainCapAveS)

Standardizing - Imputing data

set.seed(13343)

# Generating NAs to use as example
training$capAve <- training$capitalAve
selectNA <- rbinom(dim(training)[1], size = 1, prob = 0.05) == 1
training$capAve[selectNA] <- NA

# Impute and standardize
preObj <- preProcess(training[, -58], method = "knnImpute")
capAve <- predict(preObj, training[, -58])$capAve

# Standardize true values
capAveTruth <- training$capitalAve
capAveTruth <- (capAveTruth - mean(capAveTruth)) / sd(capAveTruth)

Comparing imputed to non imputed

quantile(capAve - capAveTruth)

##           0%          25%          50%          75%         100% 
## -3.654964561  0.003286070  0.003936042  0.005276720  0.988945737

quantile((capAve - capAveTruth)[selectNA])

##           0%          25%          50%          75%         100% 
## -3.654964561 -0.007730778  0.006011441  0.023683785  0.172967790

quantile((capAve - capAveTruth)[!selectNA])

##          0%         25%         50%         75%        100% 
## 0.002797249 0.003313152 0.003922613 0.005142876 0.988945737