a517dogg / getdata-035-courseproject Goto Github PK

Revision history:

library(dplyr)
revisionhistory <- data.frame(matrix(ncol = 3, nrow = 1))
names(revisionhistory) <- c("Version", "Date", "Comments")
revisionhistory$Version[1] <- "1.0"
revisionhistory$Date[1] <- "2015/12/26"
revisionhistory$Comments[1] <- "First version of document"
revisionhistory

##Introduction #####The following codebook describes the variables, the data, and any transformations or work I performed to clean up the data in the Human Activity Recognition Using Smartphones Smartphones dataset, for the Getting and Cleaning Data class in Coursera.

setwd("C:/Users/Adrian/Documents/R/coursera/getdata-035/UCI HAR Dataset/")
subjecttest <- read.table("subject_test.txt", header = FALSE)
subjecttrain <- read.table("subject_train.txt", header = FALSE)
features <- read.table("features.txt", header = FALSE)
activitylabels <- read.table("activity_labels.txt", header = FALSE)
xtest <- read.table("X_test.txt", header = FALSE)
xtrain <- read.table("X_train.txt", header = FALSE)
ytest <- read.table("y_test.txt", header = FALSE)
ytrain <- read.table("y_train.txt", header = FALSE)

####Merges the training and the test sets to create one data set.

I use rbind() to row-bind the xtrain and xtest sets, in that order. The new data is called xdata. I do the same (in the same order) for the Y data. Then I use cbind() to bind ydata and xdata. The new variable is called alldata. Task 1 is complete.

xdata <- rbind(xtrain, xtest)
ydata <- rbind(ytrain, ytest)
alldata <- cbind(ydata, xdata)

####Extracts only the measurements on the mean and standard deviation for each measurement. To do this, I want to rename the column names away from "V550" etc. to "features.txt". I want unique column names, but features has duplicate names, resulting in: "Error: found duplicated column name: etc. etc." So I use the make.names() function to make them into unique names in a new column in features. Thanks to Saeid Abolfazli in the forums for this. However because I cbinded ydata and xdata into alldata, I need to put the name for the single columm of ydata in there first. Then I can use the select() and contains() functions to get only the variables with names that contain "mean" or "std". That doesn't include the activity type from ydata, so I have to add that in too. After this, alldata contains only the ydata, the data with means, and the data with standard deviations.

features$V3 <- make.names(features$V2, unique = TRUE)
names(alldata) <- c("activitynum", features$V3)
alldata <- cbind(
  select(alldata, activitynum), 
  select(alldata, contains("mean")), 
  select(alldata, contains("std"))
  )

To do this, I want to merge the activitylabels dataframe with my data. However merge re-orders the data, so I can only use it after I've rbinded/cbinded everything together.

subject <- rbind(subjecttrain, subjecttest)
names(subject) <- "subject"
alldata <- cbind(subject, alldata)
names(activitylabels) <- c("num", "activity")
alldata <- merge(
  x = activitylabels, y = alldata, 
  by.x = "num", by.y = "activitynum", 
  all = TRUE)

I did this already in Task 2 with make.names().

To do this, I first delete the activitynum from the dataset (because it duplicates the activity name field), then group by activity and subject, then use the summarize_each() function. The new tidy data set is called "tidy."

alldata <- subset(alldata, select = -num)
alldata <- group_by(alldata, activity, subject)
tidy <- summarize_each(alldata, funs(mean))
#ungroup(alldata)

write.table(x = tidy, file = "tidydata.txt", row.names = FALSE)