README

All of the analysis is performed by the script run_analysis.R

The raw data was obtained from https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip

A detailed description of the data set can be found at http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones

0. Load the useful data files from the zip.

   • Test and training data files (X_train.txt and X_test.txt). Multiple columns containing values for each of the 561 features. Each row corresponds to a measurement on a given subject for a given activity.
   • Test and training subject files (subject_train.txt and subject_test.txt.). A single column with integers from 1 to 30 providing a link from the data to the subject.
   • Test and training activity files (y_train.txt and y_test.txt). A single column with integers from 1 to 6 providing a link from the data to the activity.
   • Activity labels file (activity_labels.txt). Two columns: the first integers from 1 to 6 and the second a description of the activity i.e. WALKING
   • Feature labels file (features.txt). Two columns: the first integers from 1 to 561 and the second a dscreption of the feature i.e. tBodyAcc-mean()-X

1. Merge the training and the test sets to create one data set.

   • The test and training subject, activity and data arrays were combined to create merged data frame containing 563 columns (subject, activity and 561 features) for all of the measurements in the test and training data.

2. Assign names to the columns.

   • The first two columns were named subject and activity. The remaining colums were named using the contents of the feature_labels data file.

3. Subset the data to keep only the fetaures with std() and mean()

   • The data frame was subsetted by selecting columns using grep/regexp to extract the columns with std() or mean() in the column name. The subject and activity columns were also kept.

4. Convert the activity column to a factor using the activity labels

   • The integer values in the activity olumn were changed into a factor variable using the contents of the activity labels data as labels.

5. Change the feature variable names such that they are readable by humans

   • A pattern for the readable names was selected. It consisted of four character strings <domain>_<measurement>_<component>_<statistic> that are defined as:
     • <domain> e.g. FrequencyDomain instead of f.
     • <measurement>: is the full spelling of the measurement e.g. BodyAcceleration instead of BodyAcc.
     • <component>: e.g. XComponent instead of X() or Magnitude instead of Mag
     • <statistic>: e.g. StandardDeviation instead of std()
   • Three of the features had the word Body repeated twice in their name. This error was found in the feature labeld file and was corrected at this stage.
   • The renaming consisted of a series of grepl/regexp calls to get the indicies and then lapply/paste0 calls to gradually build the readable column names
   • after the readable names were complete they were assigned to the feature columns of the merged and subsetted data frame.

6. Creates a second, independent tidy data set with the average of each variable for each activity and each subject.

   • The merged subset data frame was split according to both subject and activity.
   • Sapply was used to calulate the column means of the columns containing the feature variables.
   • the resulting matrix was transposed so that the measurement variables are the columns
   • the resulting matrix was written to a text file named tidy_data.txt using table.write.
     • The columns of the data contain the mean values of the features of the data set and they are labelled with the readable names.
     • The rows of the data file are not labelled but they correspond to the activity and subject <activity>.<subject> i.e. WALKING.1`.