Getting and Cleaning Data project

Description

This repository contains data files provided by http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones study and script for processing those into clean dataset. "Clean" assumes that data should be easy to deal with in future explorations.

Files

• Folder data contains unzipped the original data.
• File run_analysis.R is an R script for transforming data into tidy dataset.
• File tidy_set.csv is the result tidy dataset.
• File README.md - name speaking for itself :)

Tidy set data

For this project we take all aggregate variables from the original data set - that is, variables without postfixes [X,Y,Z]. The reason is x,y,z dimensions aren't that useful for studying overall behaviour, and they may vary across devices, subjects or even measurements - authors don't specify axes positioning. We also merge measurements data with subject id and activity name, so that future studies could dig into similarity/distinction patterns across those variables more easily. Detail description of variables in resulting file can be found in data_book.md file.

Processing data algorithm

run_analysis.R script uses function Get.Dataset(path.to.data, set.type) to obtain pre-processed test and train sets.
Then those two datasets are merged into one data.full and ordered by subject and activity name.
After that, we create tidy.set by copying the needed variables from data.full into it.
And the last step of the algorithm is to aggregate data in tidy.set by subject index and activity name, having only mean values of variables for each row.
Well, really the last step is to copy tidy.set into tidy_set.csv

function Get.Dataset(path.to.data, set.type)

The core function of run_analysis.R script is Get.Dataset(path.to.data, set.type). It takes 2 arguments: path.to.data (path to root folder of the original dataset ending with "/") and set.type ("test" or "train" depending on which dataset you wish to obtain), both being mandatory. The function returns pre-processed data.frame, containing measurements data merged with subject and activity information having proper variable names provided. Function assumes that your data folder has following files :

• ./features.txt
• ./activity_labels.txt
• ./test/X_test.txt
• ./test/y_test.txt
• ./test/subject_test.txt
• ./train/X_train.txt
• ./train/y_train.txt
• ./train/subject_train.txt

First the function creates all the needed paths to files and checks if all the files exist.
If they do, it first loads rownames and cleans them up: replaces punctuation character with ".", then merges all repeating "." into single, then trails head and tail ".".
Then function loads actual measurements from file x[test, train].txt depending on set.type value.
Then it loads subject ids and attaches them to dataset.
Then function loads activity codes and also aataches them to dataset. Then we load table with activity code - activity name values and merge it with dataset.
Column with data codes is removed from dataset, and dataset is returned.