GettingandCleaningDataCourseProject

Course Project for Coursera Getting and Cleaning Data

The result of running the run_analysis.R script is the tidy_data.txt file, which is a space delimited with header text file.

From the repository, you can run:

Rscript run_analysis.R

tidy_data.txt can be read into R:

read.table('tidy_data.txt', header=TRUE)

This was tested on R version 3.1.0 on Mac OSX 10.9.3

Note that the run_analysis.R script will look first for source files in UCI HAR Dataset/ path, and download the files from the internet if necessary

Objective:

Produce a tidy data set from the "Human Activity Recognition Using Smartphones Dataset" Version 1.0.

You should create one R script called run_analysis.R that does the following:

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

Extracts only the measurements on the mean and standard deviation for each measurement.

Uses descriptive activity names to name the activities in the data set

Appropriately labels the data set with descriptive variable names.

Creates a second, independent tidy data set with the average of each variable for each activity and each subject.

Note that since #5 is the only step that discusses the tidy data, the resulting tidy_data.txt file is the one submitted for the course project.

tidy_data.txt fulfills Tidy Data requirements

Each variable your measure should be in one column:

No single column in the tidy_data.txt file contains more than one variable. See Long Format rationale.

Each different observation of that variable should be in a different row.

For purposes of the tidy_data.txt, each observation is considered for each subject, activity, and feature. Thus, having each row uniquely identified by the subject, activity and feature.

There should be one table for each "kind" of variable

This tidy data is only one table, as we're capturing the mean by subject, activity and feature

If you have multiple tables, they should include a column in the table that allows them to be linked

Not applicable, since there is only 1 table

Rationale for long format (versus wide format)

The tidy data is in the long format, where each feature is identified as a separate row instead of being a separate column header. This allows for some scalability without dramatically altering the data format:

If we decided to include more features, it would only add more rows to the data, not add more columns
If we decided to measure the standard deviation with the mean of the features, the tidy data would only require an additional column. The wide format would nearly double the number of columns required or require a separate table to fulfill #3 of the Tidy Data requirements

Rationale for only including the mean() and std() features

The course assignment did not explicitly exclude features such as tBodyAccJerkMean, however for parsimony, the script only includes the feature that end with mean() or std(), as these features have both mean and standard deviation measures, where as the tBodyAccJerkMean has no standard deviation analogue.

Rationale for feature naming in tidy data

The feature names in the tidy data were altered as follows: