This repository contains a tidy dataset based on the dataset "Human Activity Recognition Using Smartphones Dataset Version 1.0"
This dataset itself is called mean_and_std_of_features_by_activity.txt
A list of variable names is available in features.txt.
Both these files can be generated with run_analysis.R.
This R Script assumes the original data is available in the current directory
under the name UCI HAR Dataset/.
The steps taken to derive this dataset from the original dataset are
described in CodeBook.md.
The source code of run_analysis.R also contains detailed comments
on how the script works.
The basic idea is the following. The original dataset contains two files. One with the feature names and one with the activity names. We read both these files and use the data to:
- select the features that contain mean of std values.
- replace the activity numbers with activity names.
This is done for both the test data and training data.
In a final step the data is aggregates by activity, labels are renamed, and the tidy dataset is written to disk.