dreizehnelf / coursera_data_science_gacd_week_4

Peer-graded Assignment: Getting and Cleaning Data Course Project

coursera_data_science_gacd_week_4

Peer-graded Assignment: Getting and Cleaning Data Course Project

Outline

This repository contains the necessary scripts and information to create multiple tidy datasets based on data provided by research at [http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones] and following cleaning instructions/requirements specified in the Peer-Reviewed Assignment of Week 4 of the "Getting and Cleaning Data" course on Coursera.

Steps taken to clean up the original data

(done automatically by executing the run_analysis.R script)

1. Download and extract the data from [https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip] into the root of the repository directory (will create a UCI HAR Dataset directory). This step will be skipped by the script if the directory already exists.

2. Apply a couple of denormalisations and inlines as well as some filtering (while combining the train and test datasets into one single dataset) to make the original data more accessible (see CodeBook.md or even run_analysis.R for more details on the operations performed.)

3. Save the new dataset to dist/dataset.txt (see CodeBook.md for description of the file format and on how to load the data again.)

4. Create a new dataset averaging the different features by test subject and activity (by applying the R's mean function, see run_analysis.R for the exact R code.)

5. Save the new dataset to dist/averages.txt (using the same file format)

Caveats

run_analysis.R will try to conserve bandwidth / processing time, so it will:

• NOT download the base dataset, if the UCI HAR Dataset directory already exists.
• NOT recompute the tidy dataset (before computing the averages dataset), if dist/dataset.txt already exists

So delete the appropriate files/folders if you need them re-downloaded/re-computed.