cyang-2014 / gettingandcleaningdata

Coursera R Project for "Getting And Cleaning Data" module

ReadMe for run_analysis.R

This .R script consist of 5 parts.

Part 1

• load the dplyr package
• make a lookup data table (ac_llk) wtih 2 columns - activity (number 1 - 6) and activity_nm (descriptive names) by reading the file activity_labels.txt

Part 2

Processing the files from the Train folder,

• read y_train.txt into a - this contains the list of activities performed by each volunteer (identified by subject_id)
• read subject_train.txt into s - this contains the list of subject ids associated with the activities performed
• use cbind to merge them into 1 data table (train) with 2 columns - activity and subject_id
• use mutate to add a 3rd column called data_typ that identifies this as the "Training" set
• use left join to add activity_nm column by matching activity columns in train and ac_llk. Note that left join does not re-sort the rows in the data table
• read the descriptive names for the measurement features from 'features.txt' into data table (features).
• using grep, I get the positions and full names of the Mean() and Std() features, assign them to vectors meanStdColumnPos and meanStdColumnNames
• read in the variables values from file X_train.txt into data table (v). The columns in v are not named but their names are listed in features.
• select the mean and std columns from v using the col positions from meanStdColumnPos, this is data table v_meanstd_only which I then rename its columns with extracted names in meanStdColumnsNames
• merge the train table with v_meanstd_only

The outcome of Part 2 is a data table - train (7352 x 83)

Part 3

Part 3 works like Part 2, except it processes corresponding files from the Test folder instead.

The outcome of Part 3 is a data table - test (2947 x 83)

Part 4

Taking the train and test data tables,

• use rbind_list to merge their rows into one data table - all_data (10299 x 83)

Note that the columns in all_data are now: activity, subject_id, data_typ, activity_nm and all the measurment fields related to mean() or std().

Part 5

With all_data,

• since we are not using them, throw out the data_typ, activity col by using -dat_typ, -activity in the select,
• then group_by dataset by activity_nm and subject_id,
• then summarise_each of the measurement variables by computing their mean
• write the summarised data set into external txt file - summ_data.txt in the same path as this .R script

Note that col names of the measurement variables were not changed after the summarise_each function but their values are.

The output of Part 5 is a text file - summ_data.txt (180 x 81)