Summary

This repo contains the scripts to prepare a tidy dataset from the data obtained at https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip

It will create and save as text file a dataset of average values of means and standard deviations of measurements for each activity of each subject.

Repo contents

• run_analysis.R - the main data processing script
• download_data.R - an additional script with a function to load and unpack the data if it is missing from the working directory.
• CodeBook.md - the description of variables of the resulting tidy dataset.
• README.md - explanation of data processing performed by the script.

Running the script

1. Clone this repo to your desktop
2. Launch R and change the working directory to the folder with this repo's contents.
3. If you have the UCI HAR Dataset folder or the getdata_projectfiles_UCI HAR Dataset.zip package - move or copy it to the working directory. If not - don't bother, the script will download it from the source.
4. Run the run_analysis.R script.
5. The resulting dataset will be saved as TidyDataset.txt in the working directory.

Requirements

• This script requires the plyr package. It will install and/or load it at the beginning.
• It expects the data to be located in the UCI HAR Dataset subfolder in the working directory.

Assumptions

• The structure of folders and files in UCI HAR Dataset is preserved.
• The descriptive activity names and their corresponding numeric labels are in the activity_labels.txt file. They are ordered by the numeric label.
• The names of measurements are in features.txt file. They are indexed and ordered by their indices.
  • The variables contianing the means of measurements have -mean() in their names, and those containing standard deviations of measurements have -std() in their names.
• The training data is in the train subfolder, and the test data in the test subfolder.
  • The test and train measurements are in X_train.txt and X_test.txt respectively.
  • The subject codes for the train and test datasets are in the subject_train.txt and subject_test.txt files respectively.
  • The numeric activity labels for the train and test datasets are in y_train.txt and y_test.txt respectively.
• The data in the Initial Signals subfolders is not required to complete this assignment, so the script does not use it.

Data processing

1. Store the directory and file names listed above in variables
2. Check if UCI HAR Dataset subfolder is present. If not - load the DownloadData from download_data.R which will locate and unpack or download and unpack the getdata_projectfiles_UCI HAR Dataset.zip if it is missing.
3. Process activities data:
   1. Load the train and test numeric activitiy labels and bind them in a single dataframe uciActivity.
   2. Load the descriptive activity names and use them to change the numeric labels in uciActivity to descriptive names.
4. Process subjects data:
5. Load the train and test subject codes and bind them into a single dataframe uciSubj.
6. Process measurement names:
7. Load the measurement names, extract only those containing -std() or -mean() and store these names in the character vector namesMeanSd.
8. Get row numbers of these names (which are same as their numeric indices) and store in an integer vector indexMeanSd.
9. Clean up the names: process with make.names function, then remove duplicate dots and dots from the ends. Also replace "BodyBody" in some of the names with just "Body".
10. Process measurements:
11. Load the train and test sets of measurements and bind them into a single dataframe uciData.
12. Extract a subset uciSubset selecting only the columns with indices present in indexMeanSd.
13. Label the columns of the subset with measurement names from namesMeanSd.
14. Create the tidy dataset:
15. Bind the uciSubj, uciActivity and uciSubset dataframes into a single uci dataframe.
16. Apply the ddply function to split the uci dataframe by subject and activity, calculate the means of all measurements columns and save the result into uciTidy dataframe.
17. Save uciTidy as TidyDataset.txt in the working directory.