This repository contains a R script that will create a derivative data set based on the Human Activity Recognition Using Smartphones Dataset. The data set is created by first combining the test and training subsets. From this combined data set, only feature variables for the mean and standard deviation of each measurement are selected. The activity variable is then transformed into a factor with descriptive activity labels. Finally, the data is summarized to means for each feature variable grouped by subject and activity.
The script that creates this dataset can be found in run_analysis.R. This
script assumes that the source data set has been unpacked into a directory
named UCI HAR Dataset in the current working directory. An additional
script in fetch_source_data.R can download and unpack the source data set
if necessary. Running run_analysis.R will output the derived data set in
a file named activity_summary.txt in the current working directory.
Each row represents the means of feature measurements for a given subject and activity. The first two variables on each row identify the summary group.
- subject: Identifies the subject who performed the activity. It is an integer in the range 1-30.
- activity: Identifies the type of activity associated with the feature
measurements on this row. It is a factor with six levels:
WALKING,WALKING_UPSTAIRS,WALKING_DOWNSTAIRS,SITTING,STANDING, andLAYING.
The remaining columns are means grouped by the subject and activity variables
of the feature measurements as documented in the files feature_info.txt in
the source data set. Only the variables for measurement means and standard
deviations are summarized in the derived data set. Feature variable names were
derived from the feature names by removing parenthesis characters and
replacing dashes with periods.