This file explains the analysis process.

Lets look at the data organization. There are 2 files - features.txt and activity_labels.txt that we need for both training and test data. The test and the training data are in respective folders.

First, read the features and activity_labels text files. We need the content from features file to name columns in the test and training data. We also need activity_labels to map activity id to names.

Second, read data in the train folder. Use the data from features data frame to name the columns of the test data frame. Combine the data from each of the data frames to create a training_data DF.

Repeat the procedure on test data as well. The output of this stage is test_data DF.

Combine training_data and test_data using rbind. Write the combined DF to a txt file.

Use grepl to filter out only those columns that have "mean" or "std" characters. For simplicity we should grepl "activity_id" and "subject_id" as well as we'll be using aggregation on these two attributes later.

Use factor() to convert activity levels to activity labels. Also rename activity_id to activity_name.

Use gsub and regular expressions to rename some of the columns. Some of the easy pickings are dropping (), replacing 't' with 'time', 'f' with 'frequency' etc. Also, few of the abbreviations like 'Mag', 'Gyro' can also be expanded.

Use aggregate() function to find means of all columns and group by subject_id and activity_name. Use subset() to remove the older columns from the output of aggregate() function. Write this data to 'tidy_data.txt'