The raw data were collected by Anguita et al. (2012) and are freely available at:
http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones
Anguita et al. (2012) have collected their data in experiments with a group of 30 volunteers (the subjects). Each person performed six activities wearing a smartphone on the waist. The signals measured during these activities were then subjected to additional filtering and transformations. In total, 561 features were reported per observation.
A detailed description of the original study design and of the features that were reported can be found in CODE_BOOK.md (available in this repository).
From this initial set of features, we have selected all the variables expressing the means (mean()) and the standard deviations (std()) of each variable. For each subject-activity pair, we have then calculated the average value (accross the observations) of these means and standard deviations for each feature.
#Instruction list
- Start from a folder containg the folder ./data/UCI HAR Dataset where UCI HAR Dataset contains the unzipped raw data obtained from https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip , maintaining the original folder structure.
- Run run_analysis.R
- The outputfile is average_activity_subject.txt
We have written run_analysis.R with R version 3.1.0 and we have run the script on a x86_64-w64-mingw32/x64 (64-bit) platform.
#Codebook
- The codebook containing the detailed description of all variables can be found in CODE_BOOK.md (available in this repository).