- Clone this repository
- Download the data set and extract. It should result in a
UCI HAR Datasetfolder that has all the files in the required structure. - Change current directory to the
UCI HAR Datasetfolder. - Run
Rscript <path to>/run_analysis.R - The tidy dataset should get created in the current directory as
tidy.txt
- The training and test data are available in folders named
trainandtestrespectively. - For each of these data sets:
- Measurements are present in
X_<dataset>.txtfile - Subject information is present in
subject_<dataset>.txtfile - Activity codes are present in
y_<dataset>.txtfile
- Measurements are present in
- All activity codes and their labels are in a file named
activity_labels.txt. - Names of all measurements taken are present in file
features.txtordered and indexed as they appear in theX_<dataset>.txtfiles. - All columns representing means contain
...mean()in them. - All columns representing standard deviations contain
...std()in them.
- For each of the training and test datasets,
- Read the
Xvalues - Take a subset of the columns representing only the mean and standard deviation values. Subsetting is done early on to conserve memory.
- Associate additional columns to represent activity IDs and subject IDs read from
y_<dataset>.txtandsubject_<dataset>.txtfiles respectively. - Assign column names by manipulating the measurement names in
features.txtto remove spaces and convert them to camel case.
- Read the
- Merge the training and the test sets, read as in step 1 to create one data set.
- Associate an additional column with descriptive activity names as specified in
activity_labels.txt. - Melt the dataset by specifying activity ID, name and subject ID as the only ID variables.
- Re cast the melted dataset with activity name and subject id as the only IDs and
meanas the aggregator function. - Save the resultin re-casted dataset as
tidy.txt