Practical Machine Learning Project W4
AmericasBoo
8/22/2021
Background
Using devices such as Jawbone Up, Nike FuelBand, and Fitbit it is now possible to collect a large amount of data about personal activity relatively inexpensively. These type of devices are part of the quantified self movement – a group of enthusiasts who take measurements about themselves regularly to improve their health, to find patterns in their behavior, or because they are tech geeks. One thing that people regularly do is quantify how much of a particular activity they do, but they rarely quantify how well they do it. In this project, your goal will be to use data from accelerometers on the belt, forearm, arm, and dumbell of 6 participants. They were asked to perform barbell lifts correctly and incorrectly in 5 different ways. More information is available from the website here: http://groupware.les.inf.puc-rio.br/har (see the section on the Weight Lifting Exercise Dataset).
Data Source
The training data for this project are available here:
https://d396qusza40orc.cloudfront.net/predmachlearn/pml-training.csv
The test data are available here:
https://d396qusza40orc.cloudfront.net/predmachlearn/pml-testing.csv
Intended Results
The goal of your project is to predict the manner in which they did the exercise. This is the “classe” variable in the training set. You may use any of the other variables to predict with. You should create a report describing how you built your model, how you used cross validation, what you think the expected out of sample error is, and why you made the choices you did. You will also use your prediction model to predict 20 different test cases.
Getting the Data
The training data set can be found on the following URL:
trainUrl <- "http://d396qusza40orc.cloudfront.net/predmachlearn/pml-training.csv"The testing data set can be found on the following URL:
testUrl <- "http://d396qusza40orc.cloudfront.net/predmachlearn/pml-testing.csv"Packages used for the Project
library(caret)## Loading required package: lattice## Loading required package: ggplot2library(rpart)
library(ggplot2)
library(lattice)
library(rattle)## Loading required package: tibble## Loading required package: bitops## Rattle: A free graphical interface for data science with R.
## Version 5.4.0 Copyright (c) 2006-2020 Togaware Pty Ltd.
## Type 'rattle()' to shake, rattle, and roll your data.library(randomForest)## randomForest 4.6-14## Type rfNews() to see new features/changes/bug fixes.##
## Attaching package: 'randomForest'## The following object is masked from 'package:rattle':
##
## importance## The following object is masked from 'package:ggplot2':
##
## marginlibrary(corrplot)## corrplot 0.90 loadedset.seed(1234)trainUrl <- "http://d396qusza40orc.cloudfront.net/predmachlearn/pml-training.csv"
ttestUrl <- "http://d396qusza40orc.cloudfront.net/predmachlearn/pml-testing.csv"
training <- read.csv(url(trainUrl), na.strings=c("NA","#DIV/0!",""))
testing <- read.csv(url(testUrl), na.strings=c("NA","#DIV/0!",""))Putting testset into two sets
inTrain <- createDataPartition(y=training$classe, p=0.6, list=FALSE)
myTraining <- training[inTrain, ]; myTesting <- training[-inTrain, ]
dim(myTraining); dim(myTesting)## [1] 11776 160## [1] 7846 160Cleaning the Data
Cleaning near zero variance
myDataNZV <- nearZeroVar(myTraining, saveMetrics=TRUE)myNZVvars <- names(myTraining) %in% c("new_window", "kurtosis_roll_belt", "kurtosis_picth_belt",
"kurtosis_yaw_belt", "skewness_roll_belt", "skewness_roll_belt.1", "skewness_yaw_belt",
"max_yaw_belt", "min_yaw_belt", "amplitude_yaw_belt", "avg_roll_arm", "stddev_roll_arm",
"var_roll_arm", "avg_pitch_arm", "stddev_pitch_arm", "var_pitch_arm", "avg_yaw_arm",
"stddev_yaw_arm", "var_yaw_arm", "kurtosis_roll_arm", "kurtosis_picth_arm",
"kurtosis_yaw_arm", "skewness_roll_arm", "skewness_pitch_arm", "skewness_yaw_arm",
"max_roll_arm", "min_roll_arm", "min_pitch_arm", "amplitude_roll_arm", "amplitude_pitch_arm",
"kurtosis_roll_dumbbell", "kurtosis_picth_dumbbell", "kurtosis_yaw_dumbbell", "skewness_roll_dumbbell",
"skewness_pitch_dumbbell", "skewness_yaw_dumbbell", "max_yaw_dumbbell", "min_yaw_dumbbell",
"amplitude_yaw_dumbbell", "kurtosis_roll_forearm", "kurtosis_picth_forearm", "kurtosis_yaw_forearm",
"skewness_roll_forearm", "skewness_pitch_forearm", "skewness_yaw_forearm", "max_roll_forearm",
"max_yaw_forearm", "min_roll_forearm", "min_yaw_forearm", "amplitude_roll_forearm",
"amplitude_yaw_forearm", "avg_roll_forearm", "stddev_roll_forearm", "var_roll_forearm",
"avg_pitch_forearm", "stddev_pitch_forearm", "var_pitch_forearm", "avg_yaw_forearm",
"stddev_yaw_forearm", "var_yaw_forearm")
myTraining <- myTraining[!myNZVvars]
#To check the new N?? of observations
dim(myTraining)## [1] 11776 100Remove the ID variable so that it doesn’t interfer with the algorythm
myTraining <- myTraining[c(-1)]Cleaning variables with to many NA’s
trainingV3 <- myTraining #creating another subset to iterate in loop
for(i in 1:length(myTraining)) { #for every column in the training dataset
if( sum( is.na( myTraining[, i] ) ) /nrow(myTraining) >= .6 ) { #if n?? NAs > 60% of total observations
for(j in 1:length(trainingV3)) {
if( length( grep(names(myTraining[i]), names(trainingV3)[j]) ) ==1) { #if the columns are the same:
trainingV3 <- trainingV3[ , -j] #Remove that column
}
}
}
}
dim(trainingV3)## [1] 11776 58myTraining <- trainingV3
rm(trainingV3)Now do the same transformations on the test data set
clean1 <- colnames(myTraining)
clean2 <- colnames(myTraining[, -58])
myTesting <- myTesting[clean1]
testing <- testing[clean2]
dim(myTesting)## [1] 7846 58dim(testing)## [1] 20 57We need to make the data into the same type in order to ensure proper function of the decision trees and randomForests
for (i in 1:length(testing) ) {
for(j in 1:length(myTraining)) {
if( length( grep(names(myTraining[i]), names(testing)[j]) ) ==1) {
class(testing[j]) <- class(myTraining[i])
}
}
}
testing <- rbind(myTraining[2, -58] , testing)
testing <- testing[-1,]Decision Tree
modFitA1 <- rpart(classe ~ ., data=myTraining, method="class")fancyRpartPlot(modFitA1)
predictionsA1 <- predict(modFitA1, myTesting, type = "class")## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 2150 60 7 1 0
## B 61 1260 69 64 0
## C 21 188 1269 143 4
## D 0 10 14 857 78
## E 0 0 9 221 1360
##
## Overall Statistics
##
## Accuracy : 0.879
## 95% CI : (0.871, 0.886)
## No Information Rate : 0.284
## P-Value [Acc > NIR] : <2e-16
##
## Kappa : 0.847
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.963 0.830 0.928 0.666 0.943
## Specificity 0.988 0.969 0.945 0.984 0.964
## Pos Pred Value 0.969 0.867 0.781 0.894 0.855
## Neg Pred Value 0.985 0.960 0.984 0.938 0.987
## Prevalence 0.284 0.193 0.174 0.164 0.184
## Detection Rate 0.274 0.161 0.162 0.109 0.173
## Detection Prevalence 0.283 0.185 0.207 0.122 0.203
## Balanced Accuracy 0.976 0.900 0.936 0.825 0.954Random Forest
# modFitB1 <- randomForest(classe ~. , data=myTraining)
# Predicting in-sample error:
# predictionsB1 <- predict(modFitB1, myTesting, type = "class")
# (Moment of truth) Using confusion Matrix to test results:
# confusionMatrix(predictionsB1, myTesting$classe)## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 2231 2 0 0 0
## B 1 1516 2 0 0
## C 0 0 1366 3 0
## D 0 0 0 1282 2
## E 0 0 0 1 1440
##
## Overall Statistics
##
## Accuracy : 0.999
## 95% CI : (0.997, 0.999)
## No Information Rate : 0.284
## P-Value [Acc > NIR] : <2e-16
##
## Kappa : 0.998
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 1.000 0.999 0.999 0.997 0.999
## Specificity 1.000 1.000 1.000 1.000 1.000
## Pos Pred Value 0.999 0.998 0.998 0.998 0.999
## Neg Pred Value 1.000 1.000 1.000 0.999 1.000
## Prevalence 0.284 0.193 0.174 0.164 0.184
## Detection Rate 0.284 0.193 0.174 0.163 0.184
## Detection Prevalence 0.285 0.194 0.174 0.164 0.184
## Balanced Accuracy 1.000 0.999 0.999 0.998 0.999