Run the script run_analysis.R with the data from Samsung unzipped in your working directory. Do not rename the data directory.
The scripts reads in the varibles, labels and metadata for both training and testing observations (and predictions). It merges training and testing by rows and subject/activity by column.
All variable names were cleaned up and striped of non-alpha characters.
Only variables related to means and standard deviations are selected.
The data is then aggregaed up to means grouped by activity and subject id.
see codebook.txt