GettingCleaningDataProject

This document describes the script for the Getting and Cleaning Data course project.

First, I read the X_test, y_test, and subject_test files into R X_test <- read.table("/Desktop/R/UCI HAR Dataset/test/X_test.txt", quote=""") y_test <- read.table("/Desktop/R/UCI HAR Dataset/test/y_test.txt", quote=""")

Read in the subject_test data

subject_test <- read.table("~/Desktop/R/UCI HAR Dataset/test/subject_test.txt", quote=""")

I checked the dimensions of each of the files I read in to see how they match up and determine how I should merge the files.

dim(X_test) yields 2947 rows and 561 columns dim(y_test) yields 2947 rows and 1 column

combine y_test and X_test, side by side, by columns

x_y_test <- cbind(subject_test, y_test,X_test)

change first and third headers so that there are no duplicates

names(x_y_test)[3] = "V1.2" names(x_y_test)[1] = "Subject"

change the first column to corresponding activity labels

x_y_test <- x_y_test %>% mutate(V1 = ifelse(V1 == 1, 'WALKING',V1)) x_y_test <- x_y_test %>% mutate(V1 = ifelse(V1 == 2, 'WALKING_UPSTAIRS',V1)) x_y_test <- x_y_test %>% mutate(V1 = ifelse(V1 == 3, 'WALKING_DOWNSTAIRS',V1)) x_y_test <- x_y_test %>% mutate(V1 = ifelse(V1 == 4, 'SITTING',V1)) x_y_test <- x_y_test %>% mutate(V1 = ifelse(V1 == 5, 'STANDING',V1)) x_y_test <- x_y_test %>% mutate(V1 = ifelse(V1 == 6, 'LAYING',V1))

read in the X_train and y_train files

X_train <- read.table("/Desktop/R/UCI HAR Dataset/train/X_train.txt", quote=""") y_train <- read.table("/Desktop/R/UCI HAR Dataset/train/y_train.txt", quote=""") subject_train <- read.table("~/Desktop/R/UCI HAR Dataset/train/subject_train.txt", quote=""")

combine the subject_train, y_train, and X_train files, by columns

x_y_train <- cbind(subject_train, y_train, X_train)

change first and third headers so that there are no duplicates

names(x_y_train)[3] = "V1.2" names(x_y_train)[1] = "Subject"

change the first column to corresponding activity labels

x_y_train <- x_y_train %>% mutate(V1 = ifelse(V1 == 1, 'WALKING',V1)) x_y_train <- x_y_train %>% mutate(V1 = ifelse(V1 == 2, 'WALKING_UPSTAIRS',V1)) x_y_train <- x_y_train %>% mutate(V1 = ifelse(V1 == 3, 'WALKING_DOWNSTAIRS',V1)) x_y_train <- x_y_train %>% mutate(V1 = ifelse(V1 == 4, 'SITTING',V1)) x_y_train <- x_y_train %>% mutate(V1 = ifelse(V1 == 5, 'STANDING',V1)) x_y_train <- x_y_train %>% mutate(V1 = ifelse(V1 == 6, 'LAYING',V1))

Use rbind to place x_y_test dataset above x_y_train dataset to merge the test and train datasets

x_y_dataset <- rbind(x_y_test,x_y_train)

Change second column header to descriptor "Activity"

names(x_y_dataset)[2] = "Activity"

Import features to be used as headers for new dataset

features <- read.table("~/Desktop/R/UCI HAR Dataset/features.txt", quote=""")

Convert features into character strings

featurenames <- as.character(features$V2)

Change the dataset headers to the corresponding feature names (descriptive names)

names(x_y_dataset)[3:length(names(x_y_dataset))] <- featurenames

Extract only the column headers that have either mean or standard deviation using regular expressions and store into x_y_dataset

x_y_dataset <- x_y_dataset[,grep('mean[^F]|std|Subject|Activity',names(x_y_dataset))]

Store subsetted dataset headers into new variable newfeaturenames (without activity or subject)

newfeaturenames <-featurenames[grep('mean[^F]|std|Subject|Activity',featurenames)]

Transform dataset to show mean of variables by Subject and Activity (task 5)

datasetMelt <- melt(x_y_dataset,id=c("Subject","Activity"),measure.vars=(newfeaturenames))

dcast creates the average of each activity for each subject to create new tidy dataset

summary <- dcast(datasetMelt, Subject + Activity ~ variable, mean)

dcast creates the average across all activities for each subject to create third tidy dataset

summary2 <- dcast(datasetMelt, Subject ~ variable, mean)