Prediction Assignment Writeup
Srikanth Patloori
26 April 2018
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://web.archive.org/web/20161224072740/http:/groupware.les.inf.puc-rio.br/har (see the section on the Weight Lifting Exercise Dataset).
Data
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://web.archive.org/web/20161224072740/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.
Prerequisite For Code Execution
Set working directory where files and Rmd file exists
Set seed
set.seed(1234)Load and overview of data
training <- read.csv("pml-training.csv", na.strings = c("", "NA", "#DIV/0!"))
testing <- read.csv("pml-testing.csv", na.strings = c("", "NA", "#DIV/0!"))Delete columns with all missing values and are irrelevant variables:
training <- training[, colSums(is.na(training)) == 0]
testing <- testing[, colSums(is.na(testing)) == 0]
training <- training[, -c(1:7)]
testing <- testing[, -c(1:7)]
dim(training)## [1] 19622 53dim(testing)## [1] 20 53The training data set contains 53 variables and 19622 obs. The testing data set contains 53 variables and 20 obs.
table(training$classe)##
## A B C D E
## 5580 3797 3422 3216 3607Preprocessing data In order to perform cross-validation, the training data set is partionned into 2 sets: subTraining (75%) and subTesting (25%). This will be performed using random subsampling without replacement.
library(caret)## Loading required package: lattice## Loading required package: ggplot2subsets <- createDataPartition(y = training$classe, p = 0.75, list = FALSE)
subTraining <- training[subsets, ]
subTesting <- training[-subsets, ]
dim(subTraining)## [1] 14718 53dim(subTesting)## [1] 4904 53plot(subTraining$classe
, main = "Levels of the variable classe in the subTraining data set"
, xlab = "Classe level"
, ylab = "Frequency"
)
Model Train
1. Model: Decision Treelibrary(rpart)
library(rpart.plot)
model1 <- rpart(classe ~ ., data = subTraining, method = "class")
#Testing our model performance on cross validation set.
predict1 <- predict(model1, subTesting, type = "class")
rpart.plot(model1, main = "Classification Tree", extra = 102, under = TRUE, faclen = 0)
library(e1071)
confusionMatrix(predict1, subTesting$classe)## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 1235 157 16 50 20
## B 55 568 73 80 102
## C 44 125 690 118 116
## D 41 64 50 508 38
## E 20 35 26 48 625
##
## Overall Statistics
##
## Accuracy : 0.7394
## 95% CI : (0.7269, 0.7516)
## No Information Rate : 0.2845
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.6697
## Mcnemar's Test P-Value : < 2.2e-16
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.8853 0.5985 0.8070 0.6318 0.6937
## Specificity 0.9307 0.9216 0.9005 0.9529 0.9678
## Pos Pred Value 0.8356 0.6469 0.6313 0.7247 0.8289
## Neg Pred Value 0.9533 0.9054 0.9567 0.9296 0.9335
## Prevalence 0.2845 0.1935 0.1743 0.1639 0.1837
## Detection Rate 0.2518 0.1158 0.1407 0.1036 0.1274
## Detection Prevalence 0.3014 0.1790 0.2229 0.1429 0.1538
## Balanced Accuracy 0.9080 0.7601 0.8537 0.7924 0.83072. Model: Random Forestlibrary(randomForest)## randomForest 4.6-14## Type rfNews() to see new features/changes/bug fixes.##
## Attaching package: 'randomForest'## The following object is masked from 'package:ggplot2':
##
## marginmodel2 <- randomForest(classe ~ . , data = subTraining, method = "class")
#Testing our model performance on cross validation set.
predict2 <- predict(model2, subTesting, type = "class")
confusionMatrix(predict2, subTesting$classe)## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 1395 3 0 0 0
## B 0 943 10 0 0
## C 0 3 844 5 0
## D 0 0 1 799 0
## E 0 0 0 0 901
##
## Overall Statistics
##
## Accuracy : 0.9955
## 95% CI : (0.9932, 0.9972)
## No Information Rate : 0.2845
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9943
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 1.0000 0.9937 0.9871 0.9938 1.0000
## Specificity 0.9991 0.9975 0.9980 0.9998 1.0000
## Pos Pred Value 0.9979 0.9895 0.9906 0.9988 1.0000
## Neg Pred Value 1.0000 0.9985 0.9973 0.9988 1.0000
## Prevalence 0.2845 0.1935 0.1743 0.1639 0.1837
## Detection Rate 0.2845 0.1923 0.1721 0.1629 0.1837
## Detection Prevalence 0.2851 0.1943 0.1737 0.1631 0.1837
## Balanced Accuracy 0.9996 0.9956 0.9926 0.9968 1.0000Choosing model
As expected, Random Forest algorithm performed better than Decision Trees. Accuracy for Random Forest model was 0.995 (95% CI: (0.9927, 0.9969)) compared to 0.7384 (95% CI: (0.7258, 0.7506)) for Decision Tree model. The random Forest model is choosen.
Test set prediction
predictfinal <- predict(model2, testing, type = "class")
predictfinal## 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 Eplot(predictfinal)