README file associated with the run_analysis.R R script

Assumptions and Requirements

The run_analysis.R script assumes that the UCI HAR data set has been downloaded to the local disk and that the zip file has been unzipped such that all the actual files all reside below "./UCI HAR Dataset/".

So the script will be loading file contents in order to do its work by accessing files via the path "./UCI HAR Dataset/...".

For instance, the file y_test.txt can be loaded via the following path: "./UCI HAR Dataset/test/y_test.txt"

The script has relied on basic packages typically part of standard R builds, and did not rely on plyr or other packages needed loading.

The script, when it runs successfully, produces two files that have similar contents:

"./factored_averages_txt.txt" is a text formatted file "./factored_averages_csv.csv" is a csv formatted file

Output File Description

The files have 180 rows and 81 columns, as follows:

• All columns are "tidied" in that they include column names
• The 1st column identifies the Subject, i.e., the person
• The 2nd column has the activity type (1 of 6) by character description.
• The remaining 79 columns have averages factored by subject/activity.
• The 180 rows capture the 6 activities * 30 subjects, as factors.
• Note that the file includes subjects from training and test sets.

Per the instructions, the script performs row-wise merges of various files in order to merge the training and test data set contents.

It starts off by building a column of Subject Identifiers, in order to start practicing tidiness by always naming all variables.

Then, the script uses the activity mapping file to add descriptive, character-string, activity descriptions, in addition to their coded numerical values.

The data frames above are bound column-wise to start joining things towards the output data frame with features as well as labels.

The script then performs the feature extraction functions, whereby only features that are means or standard deviations are kept. This is done by leveraging the files that list all variable names, and building intermediate data structures with the right indices for the "keeper variables" but also the corresponding names, again in the spirit of tidiness.

Finally, with the reduced data frame that only keeps features that measure means or standard deviations, we apply the aggregate() function with the proper factor list (Subject and Activity), which yields the 180 rows of averages for the 30 subjects * 6 activities.

We then write out the data frame as a .txt file and a .csv file.