Course Project for Getting and Cleaning Data course

Download the dataset from https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip and unzip it; place it at the top of the repo. It should be called "UCI HAR Dataset". This directory will be ignored by Git and will not be added to the repo.

There is only one script. Run it like so:

R CMD BATCH run_analysis.R

It produces an output file titled gcd_course_project_tidy_means.txt (which I have also uploaded separately)`.

You can read in the data like so: `tidyData <- read.table("gcd_course_project_tidy_means.txt", header=T)``

See the file Codebook.md for a codebook for this data.

Step 1 is done by merge_step1(). For each directory, test and train, the function get_data_for_subset() reads in the main data file eg X_test.txt as well as the activity number file and the subject number file (eg y_test.txtandsubject_test.txt) and uses cbind()to put them together, and gives the results column names according to features.txt. It thenrbind()`s the two sets of data.

Step 2, done by extract_mean_and_std_step2(), uses grep() to take just those observation variables that include -mean() or -std() in their name.

Step 3, done by label_activities_step3(), uses the activity_labels.txt file to replace numbers with strings (factors, really) for the Activity. It uses merge().

Step 4, done by label_variables_step4(), uses some logic to change the column names from things like tBodyAcc-mean()-Y to things like Mean.Y.axis.of.acceleration.due.to.body. It uses a couple of helper functions to do this.

Step 5 is done by tidy_dataset_step5() and it does uses split and lapply to compute the mean for each variable across each combination of (activity, subject number). This is then put back into matrix form with some dark magic.

The function do_all_the_work() runs all of these in order and writes the result out to disk.

When imported the script runs do_all_the_work() automatically. As a bonus, it keeps the produced matrix in the variable tidyData.