kampamocha / har_means

Data set with mean measurements by activity and subject of the Human Activity Response Using Smartphones data set

title	output
Getting and Cleaning Data Course Project	html_document

This file describes the script "run_analysis.R" used to summarize the "Human Activity Recognition Using Smartphones Data Set" downloaded from http://archive.ics.uci.edu/ml/machine-learning-databases/00240/UCI%20HAR%20Dataset.zip.

Input

The input is readed from the following files:

• subject_train.txt
• y_train.txt
• X_train.txt
• subject_test.txt
• y_test.txt
• X_test.txt
• features.txt
• activity_labels.txt

Process

The script reads the train and test sets of the activity recognition data, then combine both data sets to produce a unique data set selecting the columns associated with the measurements mean and standard deviation. As output the script summarizes those measurements by activity and subject. To library reshape2 is necessary to perform grouping and summarization of the variables.

The outline is the following:

• Join the train and test data selecting only the variables for mean and standard deviations estimations
• Replace activity numbers with descriptive activity names
• Summarize variable means by activity and subject

Output

The output is a tidy data set with 180 observations containing the mean of 64 measurement variables grouped by activity and subject in the file har_means.txt.