Practical Machine Learning Assignment
Prabeeti Bulani
10/11/2019
1. 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.
2. Objective
In this project, my objective is to use data from accelerometers on the belt, forearm, arm, and dumbell of 6 participants and predict the manner in which these particiapnts have prerformed these exercises. They were asked to perform barbell lifts correctly and incorrectly in 5 different ways. #Data Source & References:- #http://web.archive.org/web/20161224072740/http:/groupware.les.inf.puc-rio.br/har #(refer-Weight Lifting Exercise Data) #Training data - https://d396qusza40orc.cloudfront.net/predmachlearn/pml-training.csv #Test Set-https://d396qusza40orc.cloudfront.net/predmachlearn/pml-testing.csv
# 3. Data Loading and Exploratory Analysis
# Following library are used
library(knitr)
library(caret)## Loading required package: lattice## Loading required package: ggplot2library(rpart)
library(rpart.plot)
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(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 0.84 loadedlibrary(janitor)##
## Attaching package: 'janitor'## The following objects are masked from 'package:stats':
##
## chisq.test, fisher.testset.seed(123)
# Data Extraction
Training_url <- "https://d396qusza40orc.cloudfront.net/predmachlearn/pml-training.csv"
Testing_url <- "https://d396qusza40orc.cloudfront.net/predmachlearn/pml-testing.csv"
Training_data<-"pml-traininig.csv"
Testing_data<-"pml-testing.csv"
if(!file.exists(Training_data))
{
download.file(Training_url,destfile = Training_data,method="curl")
}
training <- read.csv(Training_data, header = TRUE, sep = ",", na.strings=c("NA","#DIV/0!",""))
dim(training)## [1] 19622 160if(!file.exists(Testing_data))
{
download.file(Testing_url,destfile = Testing_data,method="curl")
}
testing <- read.csv(Testing_data, header = TRUE, sep = ",", na.strings=c("NA","#DIV/0!",""))
dim(testing)## [1] 20 160# create a partition using caret with the training dataset on 60,40 ratio
inTraining <- createDataPartition(training$classe, p=0.6, list=FALSE)
Training_dataset <- training[inTraining, ]
Test_dataset <- training[-inTraining, ]
# Exploring the data in training and test data
dim(Training_dataset)## [1] 11776 160dim(Test_dataset)## [1] 7846 160# Training and testing dataset both have 160 variables, with records count divided as 60:40 ratio
str(Training_dataset)## 'data.frame': 11776 obs. of 160 variables:
## $ X : int 1 2 5 7 12 13 17 18 19 20 ...
## $ user_name : Factor w/ 6 levels "adelmo","carlitos",..: 2 2 2 2 2 2 2 2 2 2 ...
## $ raw_timestamp_part_1 : int 1323084231 1323084231 1323084232 1323084232 1323084232 1323084232 1323084232 1323084232 1323084232 1323084232 ...
## $ raw_timestamp_part_2 : int 788290 808298 196328 368296 528316 560359 692324 732306 740353 788335 ...
## $ cvtd_timestamp : Factor w/ 20 levels "02/12/2011 13:32",..: 9 9 9 9 9 9 9 9 9 9 ...
## $ new_window : Factor w/ 2 levels "no","yes": 1 1 1 1 1 1 1 1 1 1 ...
## $ num_window : int 11 11 12 12 12 12 12 12 12 12 ...
## $ roll_belt : num 1.41 1.41 1.48 1.42 1.43 1.42 1.51 1.55 1.57 1.59 ...
## $ pitch_belt : num 8.07 8.07 8.07 8.09 8.18 8.2 8.12 8.08 8.06 8.07 ...
## $ yaw_belt : num -94.4 -94.4 -94.4 -94.4 -94.4 -94.4 -94.4 -94.4 -94.4 -94.4 ...
## $ total_accel_belt : int 3 3 3 3 3 3 3 3 3 3 ...
## $ kurtosis_roll_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ kurtosis_picth_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ kurtosis_yaw_belt : logi NA NA NA NA NA NA ...
## $ skewness_roll_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ skewness_roll_belt.1 : num NA NA NA NA NA NA NA NA NA NA ...
## $ skewness_yaw_belt : logi NA NA NA NA NA NA ...
## $ max_roll_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ max_picth_belt : int NA NA NA NA NA NA NA NA NA NA ...
## $ max_yaw_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ min_roll_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ min_pitch_belt : int NA NA NA NA NA NA NA NA NA NA ...
## $ min_yaw_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ amplitude_roll_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ amplitude_pitch_belt : int NA NA NA NA NA NA NA NA NA NA ...
## $ amplitude_yaw_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ var_total_accel_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ avg_roll_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ stddev_roll_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ var_roll_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ avg_pitch_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ stddev_pitch_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ var_pitch_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ avg_yaw_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ stddev_yaw_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ var_yaw_belt : num NA NA NA NA NA NA NA NA NA NA ...
## $ gyros_belt_x : num 0 0.02 0.02 0.02 0.02 0.02 0 0 0 0.02 ...
## $ gyros_belt_y : num 0 0 0.02 0 0 0 0 0.02 0 0 ...
## $ gyros_belt_z : num -0.02 -0.02 -0.02 -0.02 -0.02 0 -0.02 0 -0.02 -0.02 ...
## $ accel_belt_x : int -21 -22 -21 -22 -22 -22 -21 -21 -20 -22 ...
## $ accel_belt_y : int 4 4 2 3 2 4 4 5 5 5 ...
## $ accel_belt_z : int 22 22 24 21 23 21 22 21 21 22 ...
## $ magnet_belt_x : int -3 -7 -6 -4 -2 -3 -6 1 -3 -1 ...
## $ magnet_belt_y : int 599 608 600 599 602 606 598 600 603 604 ...
## $ magnet_belt_z : int -313 -311 -302 -311 -319 -309 -317 -316 -313 -314 ...
## $ roll_arm : num -128 -128 -128 -128 -128 -128 -129 -129 -129 -129 ...
## $ pitch_arm : num 22.5 22.5 22.1 21.9 21.5 21.4 21.3 21.2 21.2 21.1 ...
## $ yaw_arm : num -161 -161 -161 -161 -161 -161 -161 -161 -161 -161 ...
## $ total_accel_arm : int 34 34 34 34 34 34 34 34 34 34 ...
## $ var_accel_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ avg_roll_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ stddev_roll_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ var_roll_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ avg_pitch_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ stddev_pitch_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ var_pitch_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ avg_yaw_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ stddev_yaw_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ var_yaw_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ gyros_arm_x : num 0 0.02 0 0 0.02 0.02 0.02 0.02 0.02 0.02 ...
## $ gyros_arm_y : num 0 -0.02 -0.03 -0.03 -0.03 -0.02 0 -0.02 -0.02 -0.02 ...
## $ gyros_arm_z : num -0.02 -0.02 0 0 0 -0.02 -0.02 -0.03 -0.02 -0.02 ...
## $ accel_arm_x : int -288 -290 -289 -289 -288 -287 -289 -288 -289 -289 ...
## $ accel_arm_y : int 109 110 111 111 111 111 110 108 109 109 ...
## $ accel_arm_z : int -123 -125 -123 -125 -123 -124 -122 -124 -122 -125 ...
## $ magnet_arm_x : int -368 -369 -374 -373 -363 -372 -371 -373 -369 -373 ...
## $ magnet_arm_y : int 337 337 337 336 343 338 337 336 340 335 ...
## $ magnet_arm_z : int 516 513 506 509 520 509 512 510 509 514 ...
## $ kurtosis_roll_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ kurtosis_picth_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ kurtosis_yaw_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ skewness_roll_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ skewness_pitch_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ skewness_yaw_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ max_roll_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ max_picth_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ max_yaw_arm : int NA NA NA NA NA NA NA NA NA NA ...
## $ min_roll_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ min_pitch_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ min_yaw_arm : int NA NA NA NA NA NA NA NA NA NA ...
## $ amplitude_roll_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ amplitude_pitch_arm : num NA NA NA NA NA NA NA NA NA NA ...
## $ amplitude_yaw_arm : int NA NA NA NA NA NA NA NA NA NA ...
## $ roll_dumbbell : num 13.1 13.1 13.4 13.1 13.1 ...
## $ pitch_dumbbell : num -70.5 -70.6 -70.4 -70.2 -70.5 ...
## $ yaw_dumbbell : num -84.9 -84.7 -84.9 -85.1 -84.9 ...
## $ kurtosis_roll_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ kurtosis_picth_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ kurtosis_yaw_dumbbell : logi NA NA NA NA NA NA ...
## $ skewness_roll_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ skewness_pitch_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ skewness_yaw_dumbbell : logi NA NA NA NA NA NA ...
## $ max_roll_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ max_picth_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ max_yaw_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ min_roll_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ min_pitch_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ min_yaw_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## $ amplitude_roll_dumbbell : num NA NA NA NA NA NA NA NA NA NA ...
## [list output truncated]# Data cleansing - cleaning the data for NA, The Near Zero variance (NZV) and ID variables
Training_nearZeroVar <- nearZeroVar(Training_dataset)
Training_dataset <- Training_dataset[, -Training_nearZeroVar]
Test_nearZeroVar <- nearZeroVar(Test_dataset)
Test_dataset <- Test_dataset[, -Test_nearZeroVar]
label <- apply(Training_dataset, 2, function(x) mean(is.na(x))) > 0.95
Training_dataset <- Training_dataset[, -which(label, label == FALSE)]
label_test <- apply(Test_dataset, 2, function(x) mean(is.na(x))) > 0.95
Test_dataset <- Test_dataset[, -which(label_test, label_test == FALSE)]
Training_dataset <- Training_dataset[ , -(1:5)]
Test_dataset <- Test_dataset[ , -(1:5)]
dim(Training_dataset)## [1] 11776 54dim(Test_dataset)## [1] 7846 54# Training and testing dataset have 54 and 54 variables respectively after cleansing.4.Predictive Modelling Processes
4.1 Decision Tree Model
# DEcision tree build
set.seed(123)
model_decisionTree <- rpart(classe ~ ., Training_dataset, method="class")
fancyRpartPlot(model_decisionTree)## Warning: labs do not fit even at cex 0.15, there may be some overplotting
# Prediction with decision tree model
predict_DT<- predict(model_decisionTree, Test_dataset, type = "class")
confusionMatrix_DT <-confusionMatrix(predict_DT, Test_dataset$classe)
confusionMatrix_DT## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 1955 122 3 25 8
## B 124 1147 59 43 80
## C 0 76 1100 39 4
## D 134 130 177 1052 109
## E 19 43 29 127 1241
##
## Overall Statistics
##
## Accuracy : 0.8278
## 95% CI : (0.8193, 0.8361)
## No Information Rate : 0.2845
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.7828
##
## Mcnemar's Test P-Value : < 2.2e-16
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.8759 0.7556 0.8041 0.8180 0.8606
## Specificity 0.9719 0.9516 0.9816 0.9162 0.9660
## Pos Pred Value 0.9252 0.7894 0.9024 0.6567 0.8506
## Neg Pred Value 0.9517 0.9420 0.9596 0.9625 0.9685
## Prevalence 0.2845 0.1935 0.1744 0.1639 0.1838
## Detection Rate 0.2492 0.1462 0.1402 0.1341 0.1582
## Detection Prevalence 0.2693 0.1852 0.1554 0.2042 0.1860
## Balanced Accuracy 0.9239 0.8536 0.8929 0.8671 0.9133# plot confusion matrix results
plot(confusionMatrix_DT$table, col = confusionMatrix_DT$byClass,
main = paste("Decision Tree - Accuracy =",
round(confusionMatrix_DT$overall['Accuracy'], 4)))
4.2 Randon Forest Model
# Random Forest build
set.seed(123)
train_ctrl <- trainControl(method = "cv", classProbs=TRUE,savePredictions=TRUE,allowParallel=TRUE, number = 10)
model_randomForest <- randomForest(classe ~ ., data = Training_dataset, method = "rf", importance = T, trControl = train_ctrl,na.action = na.exclude)
# Prediction with random forest model
predict_RF<- predict(model_randomForest, Test_dataset, type = "class")
confusionMatrix_RF <-confusionMatrix(predict_RF, Test_dataset$classe)
confusionMatrix_RF## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 2231 2 0 0 0
## B 0 1515 11 0 0
## C 0 1 1357 11 0
## D 0 0 0 1274 2
## E 1 0 0 1 1440
##
## Overall Statistics
##
## Accuracy : 0.9963
## 95% CI : (0.9947, 0.9975)
## No Information Rate : 0.2845
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9953
##
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.9996 0.9980 0.9920 0.9907 0.9986
## Specificity 0.9996 0.9983 0.9981 0.9997 0.9997
## Pos Pred Value 0.9991 0.9928 0.9912 0.9984 0.9986
## Neg Pred Value 0.9998 0.9995 0.9983 0.9982 0.9997
## Prevalence 0.2845 0.1935 0.1744 0.1639 0.1838
## Detection Rate 0.2843 0.1931 0.1730 0.1624 0.1835
## Detection Prevalence 0.2846 0.1945 0.1745 0.1626 0.1838
## Balanced Accuracy 0.9996 0.9981 0.9951 0.9952 0.9992# plot confusion matrix results for random forest
plot(model_randomForest)
4.3 Generalised Boosting Model
# Generalised Boosting Model build
set.seed(123)
gbm_control <- trainControl(method = "repeatedcv", number = 5, repeats =1,verboseIter = FALSE)
model_Boosting <- train(classe ~ ., method = "gbm",data= Training_dataset,
verbose = FALSE,
trControl = gbm_control)
model_Boosting## Stochastic Gradient Boosting
##
## 11776 samples
## 53 predictor
## 5 classes: 'A', 'B', 'C', 'D', 'E'
##
## No pre-processing
## Resampling: Cross-Validated (5 fold, repeated 1 times)
## Summary of sample sizes: 9420, 9421, 9420, 9421, 9422
## Resampling results across tuning parameters:
##
## interaction.depth n.trees Accuracy Kappa
## 1 50 0.7572160 0.6919090
## 1 100 0.8325396 0.7880080
## 1 150 0.8694789 0.8348284
## 2 50 0.8828099 0.8516050
## 2 100 0.9369047 0.9201733
## 2 150 0.9617859 0.9516436
## 3 50 0.9290921 0.9102227
## 3 100 0.9700235 0.9620708
## 3 150 0.9864979 0.9829213
##
## Tuning parameter 'shrinkage' was held constant at a value of 0.1
##
## Tuning parameter 'n.minobsinnode' was held constant at a value of 10
## Accuracy was used to select the optimal model using the largest value.
## The final values used for the model were n.trees = 150,
## interaction.depth = 3, shrinkage = 0.1 and n.minobsinnode = 10.plot(model_Boosting)
# Prediction with Generalised Boosting Model model
predict_GBM<- predict(model_Boosting, Test_dataset)
confusionMatrix_GBM <-confusionMatrix(predict_GBM, Test_dataset$classe)
confusionMatrix_GBM## Confusion Matrix and Statistics
##
## Reference
## Prediction A B C D E
## A 2227 16 0 1 1
## B 5 1481 14 5 4
## C 0 16 1349 14 8
## D 0 5 5 1262 10
## E 0 0 0 4 1419
##
## Overall Statistics
##
## Accuracy : 0.9862
## 95% CI : (0.9834, 0.9887)
## No Information Rate : 0.2845
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9826
##
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: A Class: B Class: C Class: D Class: E
## Sensitivity 0.9978 0.9756 0.9861 0.9813 0.9840
## Specificity 0.9968 0.9956 0.9941 0.9970 0.9994
## Pos Pred Value 0.9920 0.9814 0.9726 0.9844 0.9972
## Neg Pred Value 0.9991 0.9942 0.9971 0.9963 0.9964
## Prevalence 0.2845 0.1935 0.1744 0.1639 0.1838
## Detection Rate 0.2838 0.1888 0.1719 0.1608 0.1809
## Detection Prevalence 0.2861 0.1923 0.1768 0.1634 0.1814
## Balanced Accuracy 0.9973 0.9856 0.9901 0.9891 0.9917Conclusion
#Accuracy obtained from 3 models are as below:- #1. Decision Tree Model -0.8278 #2. Random Forest Model - 0.9963 #3. Boosting Model - 0.9862
Out of these, accuracy is best for random forest model , so Random Forest model is applied to predict the 20 quiz results
predict_RF<- predict(model_randomForest, testing)
predict_RF## 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 EI have used assumption in processing data is to limit dataset for attributes that are non-zero in the Training and tEsting dataset. I haved faced lots of error about missing attributes in Testing data but available in Training dataset but this got tackeled by using this assumption.