Getting and Cleaning Data - Week 4 Assignment

Content of the repository

• README.md : this file.

• CodeBook.md : description of the feature.

• features.txt : list of features.

• run_analysis.R : R script analyzing the original dataset.

This week's script does the following.

1. Merges the training and the test sets to create one data set from the several files in the original dataset.

We use dplyr bind_rows() and bind_cols to merge the various datasets together.

First we start by reading the activities and the features. Then we load the train/test "inputs" (X_train.txt and X_test.txt), as well as the train/test "outputs" (Y_train.txt and Y_test.txt).

We merge the train data together, and the test data together. Then we add a dataset column to each that specifies whether a row is a training row or a test row.

The we use bind_cols() to merge the train and test together.

=> dataframe1

2. Extracts only the measurements on the mean and standard deviation for each measurement.

Here dplyr's select function comes in handy, using a regexp inside a match(), we manage to only keep the columns that contain mean or standard deviation quantities.

=> dataframe2

3. Uses descriptive activity names to name the activities in the data set

A join, or more specifically an inner join is necessary here. Before hand, we loaded the activities dataframe with two columns:

• activityid

• activitylabel

We only need to call inner_join() to join dataframe2 and the activities dataframe, we join by activityid column as it is common to both dataframes. This will add a column activitylabel to dataframe2.

=> dataframe3

4. Appropriately labels the data set with descriptive variable names.

This step was done right after reading the data by assigning the descriptive set of feature names (read from the file features.txt) to name(dataset) for each one of the the training and test set.

=> dataframe4

5. Create a tidy data set with the average of each variable for each activity and each subject.

Using chaining capabilities of dplyr we group_by the dataframe4, using the variables subjectid and activitylabel (and dataset) and then summarize the grouped structure, by applying the mean function to all the non-grouping variables, which correspond to the feature columns. They happen to be all of type double, so using summarize_if(is.double, ...) does the trick.

=> dataframe5

Result

Now we have a dataset where:

• each row represents one observation of one subject in one activity.

• Each column represents exactly one variable.

• The observational units are in one table.

=> The result is a happy tidy dataset.