The script run_analysis.R clean the data. It does the following:

• The training and testing dataset are first read into two variables, train and test, respectively
• The training & testing dataset are then merged using rbind into a single variable named all

Notes:

• In the first part of the code, I only merged X_train & X_test. y_train and y_test, subject_train, subject_test are merged into all later in part 3 for coding convenience.

• First the list of all feature names are read into variable features
• Then the features with mean and standard deviation are found using regular expression. The column indices are stored in variable good_columns
• Variable all is subsetted with good_columns

Notes:

• In here I extracted 66 columns. There are columns "meanFreq" that I purposely excluded. Only columns that has "mean()" and "std()" in its name are included

• First the English names of the activites are read into variable activity_labels.
• Then activity_labels are used to labels y_train and y_test, and then merged into all
• Next, subject are read and merged into subject_all
• Finally, subject_all is merged into all. After this step, all contains all the information that we need to merge

• The average are calculated using function aggregate. The result is stored in variable tidy which will be then stored into a file named "tidy_data.txt"
• tidy has 180 rows. This is because: There are 30 subject and 6 activities, thus there are 30*6 = 180 averages for each variable.