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, we 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.The data consists of a Training data and a Test data
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.
trainD<-read.csv ("pml-training.csv",header = TRUE)
validD<-read.csv ("pml-testing.csv",header = TRUE)
dim(trainD)
## [1] 19622 160
dim(validD)
## [1] 20 160
some useful packages:
library(caret)
## Loading required package: lattice
## Loading required package: ggplot2
library(corrplot)
## corrplot 0.84 loaded
Note along the cleaning process we display the dimension of the reduced dataset Here,
# Removing the variables that contains missing values.
trainD<- trainD[, colSums(is.na(trainD)) == 0]
validD <- validD[, colSums(is.na(validD)) == 0]
dim(trainD)
## [1] 19622 93
dim(validD)
## [1] 20 60
and
# removing the first seven variables as they have little impact on the outcome classe
trainD <- trainD[, -c(1:7)]
validD <- validD[, -c(1:7)]
dim(trainD)
## [1] 19622 86
dim(validD)
## [1] 20 53
Preparing the datasets for prediction:
set.seed(123)
R_training <- createDataPartition(y=trainD$classe, p=0.7, list=F)
training <- trainD[R_training, ]
testing <- trainD[-R_training, ]
Further, cleaning the data by removing the variables that are nearly zero variance
# remove variables with nearly zero variance
nzv <- nearZeroVar(training)
training <- training[, -nzv]
testing <- testing[, -nzv]
dim(training)
## [1] 13737 53
dim(testing)
## [1] 5885 53
After this cleaning we are down now to 53 variables
Begin with buildilding a model using classification trees.
Here is a classification tree:
library(rattle)
## Rattle: A free graphical interface for data science with R.
## Version 5.2.0 Copyright (c) 2006-2018 Togaware Pty Ltd.
## Type 'rattle()' to shake, rattle, and roll your data.
library(rpart)
set.seed(345)
modelfit_tree <- rpart(classe ~ ., data=training, method="class")
fancyRpartPlot(modelfit_tree)
We then validate the model “modelfit_tree” on the testData to find out how well it performs by looking at the accuracy variable:
predict_tree <- predict(modelfit_tree, testing, type = "class")
cm_tree <- confusionMatrix(predict_tree, testing$classe)
cm_tree
## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 1424 184 52 56 25
## B 26 667 54 48 70
## C 33 107 701 53 52
## D 167 106 193 735 157
## E 24 75 26 72 778
##
## Overall Statistics
##
## Accuracy : 0.7315
## 95% CI : (0.72, 0.7428)
## No Information Rate : 0.2845
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.6606
##
## Mcnemar's Test P-Value : < 2.2e-16
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.8507 0.5856 0.6832 0.7624 0.7190
## Specificity 0.9247 0.9583 0.9496 0.8734 0.9590
## Pos Pred Value 0.8179 0.7711 0.7410 0.5412 0.7979
## Neg Pred Value 0.9397 0.9060 0.9342 0.9494 0.9381
## Prevalence 0.2845 0.1935 0.1743 0.1638 0.1839
## Detection Rate 0.2420 0.1133 0.1191 0.1249 0.1322
## Detection Prevalence 0.2958 0.1470 0.1607 0.2308 0.1657
## Balanced Accuracy 0.8877 0.7719 0.8164 0.8179 0.8390
We see that the accuracy rate of the model is low: 0.7315 and therefore the out-of-sample-error is about 0.27 which is considerable.
Random Forests model
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':
##
## margin
Fitting the model
controlRF <- trainControl(method="cv", number=3, verboseIter=FALSE)
modRF1 <- train(classe ~ ., data=training, method="rf", trControl=controlRF)
modRF1$finalModel
##
## Call:
## randomForest(x = x, y = y, mtry = param$mtry)
## Type of random forest: classification
## Number of trees: 500
## No. of variables tried at each split: 27
##
## OOB estimate of error rate: 0.67%
## Confusion matrix:
## A B C D E class.error
## A 3901 3 1 0 1 0.001280082
## B 23 2629 5 1 0 0.010910459
## C 0 12 2379 5 0 0.007095159
## D 0 0 27 2224 1 0.012433393
## E 0 1 5 7 2512 0.005148515
I see that it decided to use 500 trees and try 27 variables at each split.
Model Evaluation:
predictRF1 <- predict(modRF1, newdata=testing)
cmrf <- confusionMatrix(predictRF1, testing$classe)
cmrf
## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 1671 8 0 0 0
## B 3 1130 8 0 0
## C 0 1 1015 19 2
## D 0 0 3 944 1
## E 0 0 0 1 1079
##
## Overall Statistics
##
## Accuracy : 0.9922
## 95% CI : (0.9896, 0.9943)
## No Information Rate : 0.2845
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9901
##
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.9982 0.9921 0.9893 0.9793 0.9972
## Specificity 0.9981 0.9977 0.9955 0.9992 0.9998
## Pos Pred Value 0.9952 0.9904 0.9788 0.9958 0.9991
## Neg Pred Value 0.9993 0.9981 0.9977 0.9959 0.9994
## Prevalence 0.2845 0.1935 0.1743 0.1638 0.1839
## Detection Rate 0.2839 0.1920 0.1725 0.1604 0.1833
## Detection Prevalence 0.2853 0.1939 0.1762 0.1611 0.1835
## Balanced Accuracy 0.9982 0.9949 0.9924 0.9892 0.9985
The accuracy rate using the random forest is very high: Accuracy : 0.9922 and therefore the out-of-sample-error is equal to 0.0078.This is an excellent result, so rather than trying additional algorithms, I will use Random Forests to predict on the test set.
All that is left is to use this model to predict the classes of the validation data:
pred_val <- predict(modRF1, validD)
pred_val
## [1] 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 E
The Results output will be used to answer the “Course Project Prediction Quiz”