This project focuses on cleaning Samsung smartphone accelerometer data and shaping it into a tidy data set useful for analysis.
- Load all data sets (feature, subject, training and test data)
- Create the full set of data by merging training and test observation rows
- Merge the training and test activity id rows
- Merge the training and test subject id rows
- Filter the feature names to just mean/std observations, then filter down full data set to reduce to the core data frame required
- Append the activity ids and subject ids to the core data frame
- Convert the activity ids to descriptive labels
- Convert the data set into a long data frame by gathering all feature variables
- (optional) Separate the feature variables into the constituent components of feature, measure and axis
- Group and summarise the result set on activity, subject id and variable