| title | author | date | output |
|---|---|---|---|
README - run_analysis.R |
Ian |
Sunday, March 22, 2015 |
html_document |
This repository contains a R script produced for the course project of the Getting and Cleaning Data course of the Data Science Specialization provided by Coursera:
https://www.coursera.org/course/getdata
The script processes the data from a research study of Human Activity Using Smartphones to produce a 'tidy' rectangular dataset.
The original datasets, together with documentation of the study and of these original data, are available here:
http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones
The script assumes that the data distributed from this site has been unzipped in the working directory preserving its directory structure.
The output dataset is written to a text file: "Getting_Cleaning_Project_Data.txt".
It comprises a single rectangular data table made up of 72 variables and 180 records, together with a header record that contains variable names.
The data fields are separated by spaces.
The data comprise 70 different measurements on each of 30 subjects while they were engaged in each of 6 activities.
The script produces a "wide" format dataset that comprises one record for each activity for each subject that contains subject and activity identifiers and all of the measurements.
A more detailed codebook is included in this repository.
The script was written using R 3.1.2 under Windows 7 SP1.
It performs 5 tasks successively:
- Merging the training and the test datasets to create one data set. This step also prefixes the measurements with Subject and Activity identifiers obtained from the subject and y tables.
- Extracting the mean and standard deviation features of each measurement.
- Converting the Activity variable to a factor with descriptive activity names obtained from the activity_labels table.
- Setting the names of the measurement variables to be descriptive variable names obtained from the features table. (At this point the script writes the disaggregated data to a file).
- Creating and writing to file an aggregated dataset containing the averages of the measures calulated across each Activity by Subject pairing. This dataset is in wide format (i.e. a file with the averages of the whole series of mean measurements for each Activity by Subject pairing on a single record).
Further comments on detailed aspects of some of these steps are included in the run_analysis.R script file.