yaksha13 / cleaningdataproj Goto Github PK

The Project's objective was to take raw data collected from accelerometers from the Samsung Galaxy S smartphone by subjects and prepare a clean tidy data set

This project contains the following:

• README.md : General description of the project
• CodeBook.md: Describes the variables and measurements in the tidy data set
• run_analysis.R: Cleaning script for preparing the tidy dataset from the raw data

The initial data for the cleaning exercise was obtained from: http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones and https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip

You can find the tidy data set uploaded here: https://s3.amazonaws.com/coursera-uploads/user-5148c4bb98a5b56203474727/973500/asst-3/53114f10ebec11e4a9166b9aa399b77f.txt

Descriptions of the observations and the measurements in the tidy data set have been described in CodeBook.md

The cleaning script and resultant dataset (linked above) follows the following tidy data set principles:

• Each variable measured is in ONE column
• Each observation for that variable is in a different row
• There is only ONE table for each 'kind' of variable

The script performs the following major functions:

• Take data in the train and test sets (for subjects, activity and features) and merge them into a master data set
• Provide the appropriate labels that are descriptive for the variable names. Here the names provided in features.txt of the Raw data was used
• Create a smaller dataset that only has the mean() and std() feature measurements for the raw data
• After creating the smaller dataset, replace the factor style values for the Activity variable with more descriptive alternatives
  • Again, activity names provided in activity_labels.txt of the Raw Data was used
• Group the multiple observations (and feature measurements) for each subject and each activity, substituting it with the average over all those measurements
• Finally, melt the "wide"-style tidy data set into a "narrow-tall" style dataset by combining all the different feature measurements (now averaged over multiple subject/ activity observations) into a single "feature" column and its corresponding measurement value in "featureValue" column
• write the final tidy data out to "tidydataset.txt"