This project contains my submission for the Project assignment of the Getting and Cleaning Data class on Coursera.
The files in this project are:
README.md: this file. Explains how all of the scripts workCodeBook.md: describes the variables, the data, and any transformations or work that was performed to clean up the datarun_analysis.R: code, commented inside and below in this file
As requested in the assignment, download the data file here: https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip.
Unzip it: this will create a folder named UCI HAR Dataset that contains the data files.
Download the script run_analysis.R in the same folder (at the same level) as UCI HAR Dataset.
In R, source it:
- make the directo
6F59
ry that constains the script the working directory (using the
cwdcommand) - execute
source("run_analysis.R") - this will launch the script
The script will generate in the directory 2 files:
questions_1_4.txt: contains the output of the sequence of the questions 1 to 4question_5.txt: contains the output of the question 5
The loadRawData function takes care of loading all the relevant data (the X, y and subject files for resp. the train and test file collections), and bind them in a single data frame.
For the test files, the 3 X, y and subject files are actually loaded by the same loadTable function, and then they are coloumn bound. This is done in the loadAndBindfunction that is also used to load the train data, given that the files organization is the same in both data sets.
After that, the 2 resulting data frames are row-bound.
At last, the column names are loaded from the features file, and completed so that appropriate column names are assigned to the global data frame.
This is done using the subset R function, and a regular expression to keep only columns which have a name containing subject, mean, std or equaling y:
data <- data[,grepl("(subject)|(mean)|(std)|(^y$)", names(data))]
This is done in 3 operations in the setDescriptiveActivityNames function:
- load the
activity_labels.txtin a data frame, and name its columns resp.yandactivity - execute an R
mergeon theycolumn between this data frame and the actual data frame we prepared before (both have aycolumn); this builds a join. - remove the now useless
ycolumn in the resulting data frame
We use a simple transformation to rename the variables that look like tBodyAcc-mean()-X in to something like timeBodyAcc.mean.X:
- replace
tprefixes byTimeandfprefixes byFreq - replace
-separators by.(seems to be a more common convention in R) - remove the parenthesis that do not really add any useful information to the reader
This is done in the tidyColNames function using regular expressions and the R gsub function.
This is done using the plyr and reshape2 libraries that may need to be installed (using install.packages)
- first we melt all the features column, keeping the Subject and Activity columns as "indexes": this creates a table with 3 columns (Subject, Activity, and a "variable")
- the we use
ddplyto apply ameancomputation on the "variable column", grouping by Subject and Activity - at last we set nicer column names and we save the file