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).
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
The data for this project come from this source: http://groupware.les.inf.puc-rio.br/har. If you use the document you create for this class for any purpose please cite them as they have been very generous in allowing their data to be used for this kind of assignment.
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 and loading the data
# Loading Data
# In this part we are going to load the dataset, and attached it to the environment
library(caret)## Loading required package: lattice## Loading required package: ggplot2library(mlbench)
library(randomForest)## randomForest 4.6-12## Type rfNews() to see new features/changes/bug fixes.##
## Attaching package: 'randomForest'## The following object is masked from 'package:ggplot2':
##
## marginlibrary(foreach)
library(doParallel)## Loading required package: iterators## Loading required package: parallellibrary(Hmisc)## Loading required package: survival##
## Attaching package: 'survival'## The following object is masked from 'package:caret':
##
## cluster## Loading required package: Formula##
## Attaching package: 'Hmisc'## The following object is masked from 'package:randomForest':
##
## combine## The following objects are masked from 'package:base':
##
## format.pval, round.POSIXt, trunc.POSIXt, unitsset.seed(39)
trainUrl <-"https://d396qusza40orc.cloudfront.net/predmachlearn/pml-training.csv"
testUrl <- "https://d396qusza40orc.cloudfront.net/predmachlearn/pml-testing.csv"
ptrain <- read.csv(url(trainUrl), na.strings=c("NA","#DIV/0!",""))
ptest <- read.csv(url(testUrl), na.strings=c("NA","#DIV/0!",""))Here we are going to remove all the NA from the variables. We also going to split data in two, we are going to use 75% of the data for training and another 25 for testing.
rmNaTrain <- ptrain[, apply(ptrain, 2, function(x) !any(is.na(x)))]
training <-rmNaTrain
dim(training)## [1] 19622 60# cleaning variables
clVariables <-rmNaTrain[,-c(1:8)]
training <- clVariables
dim(training)## [1] 19622 52cleanpmTest <-ptest[,names(training[,-52])]
testing <- cleanpmTest
dim(testing)## [1] 20 51# Splitting Data
library(caret)
set.seed(39)
inTrain <- createDataPartition(y=training$classe, p=0.75, list=F)
ptrain1 <- training[inTrain, ]
ptrain2 <- training[-inTrain, ]
dim(ptrain1)## [1] 14718 52dim(ptrain2)## [1] 4904 52Now we will take a look at the data set a few different ways
# Dimensions of the Dataset
# Level
levels(ptrain1$classe)## [1] "A" "B" "C" "D" "E"We will use random forest and Gradient Boosting for comparison on which algorithms best describes the data.
library(randomForest)
library(caret)
set.seed(13333)
control <- trainControl(method = "cv", number = 5, allowParallel = T, verbose=T)
rf.formula = randomForest(classe~., data=ptrain1, method="rf",trControl=control,verbose=F)mpredict <- predict(rf.formula, newdata=ptrain2)
confusionMatrix(mpredict, ptrain2$classe)## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 1395 2 0 0 0
## B 0 947 7 0 0
## C 0 0 847 16 1
## D 0 0 1 788 2
## E 0 0 0 0 898
##
## Overall Statistics
##
## Accuracy : 0.9941
## 95% CI : (0.9915, 0.996)
## No Information Rate : 0.2845
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9925
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 1.0000 0.9979 0.9906 0.9801 0.9967
## Specificity 0.9994 0.9982 0.9958 0.9993 1.0000
## Pos Pred Value 0.9986 0.9927 0.9803 0.9962 1.0000
## Neg Pred Value 1.0000 0.9995 0.9980 0.9961 0.9993
## Prevalence 0.2845 0.1935 0.1743 0.1639 0.1837
## Detection Rate 0.2845 0.1931 0.1727 0.1607 0.1831
## Detection Prevalence 0.2849 0.1945 0.1762 0.1613 0.1831
## Balanced Accuracy 0.9997 0.9981 0.9932 0.9897 0.9983predSec <- predict(rf.formula,newdata=testing)
predSec## 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
## B A B A A E D B A A B C B A E E A B B B
## Levels: A B C D Econtrol <- trainControl(method = "cv", number = 5, allowParallel = T, verbose=T)
fit.gbm <-train(classe~.,data=ptrain1, method="gbm", trControl=control, verbose=F)## Loading required package: gbm## Loading required package: splines## Loaded gbm 2.1.3## Loading required package: plyr##
## Attaching package: 'plyr'## The following objects are masked from 'package:Hmisc':
##
## is.discrete, summarize## + Fold1: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## - Fold1: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## + Fold1: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## - Fold1: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## + Fold1: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## - Fold1: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## + Fold2: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## - Fold2: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## + Fold2: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## - Fold2: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## + Fold2: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## - Fold2: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## + Fold3: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## - Fold3: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## + Fold3: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## - Fold3: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## + Fold3: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## - Fold3: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## + Fold4: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## - Fold4: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## + Fold4: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## - Fold4: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## + Fold4: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## - Fold4: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## + Fold5: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## - Fold5: shrinkage=0.1, interaction.depth=1, n.minobsinnode=10, n.trees=150
## + Fold5: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## - Fold5: shrinkage=0.1, interaction.depth=2, n.minobsinnode=10, n.trees=150
## + Fold5: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## - Fold5: shrinkage=0.1, interaction.depth=3, n.minobsinnode=10, n.trees=150
## Aggregating results
## Selecting tuning parameters
## Fitting n.trees = 150, interaction.depth = 3, shrinkage = 0.1, n.minobsinnode = 10 on full training setfit.gbm$finalModel## A gradient boosted model with multinomial loss function.
## 150 iterations were performed.
## There were 51 predictors of which 41 had non-zero influence.class(fit.gbm)## [1] "train" "train.formula"gbmPred <- predict(fit.gbm, newdata=ptrain2)
confusionMatrix(gbmPred,ptrain2$classe)## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 1365 28 0 1 0
## B 22 889 30 3 4
## C 6 27 812 28 11
## D 2 3 13 761 10
## E 0 2 0 11 876
##
## Overall Statistics
##
## Accuracy : 0.959
## 95% CI : (0.9531, 0.9644)
## No Information Rate : 0.2845
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9482
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.9785 0.9368 0.9497 0.9465 0.9723
## Specificity 0.9917 0.9851 0.9822 0.9932 0.9968
## Pos Pred Value 0.9792 0.9378 0.9186 0.9645 0.9854
## Neg Pred Value 0.9915 0.9848 0.9893 0.9896 0.9938
## Prevalence 0.2845 0.1935 0.1743 0.1639 0.1837
## Detection Rate 0.2783 0.1813 0.1656 0.1552 0.1786
## Detection Prevalence 0.2843 0.1933 0.1803 0.1609 0.1813
## Balanced Accuracy 0.9851 0.9609 0.9660 0.9698 0.9845predtrain <-predict(fit.gbm,newdata=ptrain1)
confusionMatrix(predtrain, ptrain1$classe)## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 4141 81 0 0 2
## B 33 2711 52 9 14
## C 7 52 2485 57 16
## D 3 1 25 2324 26
## E 1 3 5 22 2648
##
## Overall Statistics
##
## Accuracy : 0.9722
## 95% CI : (0.9694, 0.9748)
## No Information Rate : 0.2843
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9648
## Mcnemar's Test P-Value : 1.141e-09
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.9895 0.9519 0.9681 0.9635 0.9786
## Specificity 0.9921 0.9909 0.9891 0.9955 0.9974
## Pos Pred Value 0.9804 0.9617 0.9496 0.9769 0.9884
## Neg Pred Value 0.9958 0.9885 0.9932 0.9929 0.9952
## Prevalence 0.2843 0.1935 0.1744 0.1639 0.1839
## Detection Rate 0.2814 0.1842 0.1688 0.1579 0.1799
## Detection Prevalence 0.2870 0.1915 0.1778 0.1616 0.1820
## Balanced Accuracy 0.9908 0.9714 0.9786 0.9795 0.9880preditrain <- predict(fit.gbm, newdata=ptrain1)
confusionMatrix(preditrain, ptrain1$classe)## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 4141 81 0 0 2
## B 33 2711 52 9 14
## C 7 52 2485 57 16
## D 3 1 25 2324 26
## E 1 3 5 22 2648
##
## Overall Statistics
##
## Accuracy : 0.9722
## 95% CI : (0.9694, 0.9748)
## No Information Rate : 0.2843
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9648
## Mcnemar's Test P-Value : 1.141e-09
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.9895 0.9519 0.9681 0.9635 0.9786
## Specificity 0.9921 0.9909 0.9891 0.9955 0.9974
## Pos Pred Value 0.9804 0.9617 0.9496 0.9769 0.9884
## Neg Pred Value 0.9958 0.9885 0.9932 0.9929 0.9952
## Prevalence 0.2843 0.1935 0.1744 0.1639 0.1839
## Detection Rate 0.2814 0.1842 0.1688 0.1579 0.1799
## Detection Prevalence 0.2870 0.1915 0.1778 0.1616 0.1820
## Balanced Accuracy 0.9908 0.9714 0.9786 0.9795 0.9880getwd()## [1] "E:/DScience/Practical_Machine_Learning/Project"pml_write_filles = function(x){
n = length(x)
for(i in 1:n){
filename=paste0("problem_id_",i,"txt")
write.table(x[i],file = filename, quote = FALSE, row.names = FALSE, col.names = FALSE)
}
}
pml_write_filles## function(x){
## n = length(x)
## for(i in 1:n){
## filename=paste0("problem_id_",i,"txt")
##
## write.table(x[i],file = filename, quote = FALSE, row.names = FALSE, col.names = FALSE)
##
## }
## }