Coursera Data Science Specialization

Getting and Cleaning Data

WARNING: If the file you are seeing has the .Rmd extension you should go to README.md

NOTE: This repo does not contain the raw data, it has a script that downloads and unzips the data into the current working directory

This repository holds a project that tries to fulfill the assinged objectives on the Coursera Data Science Specialization - Getting and Cleaning Data course from the John Hopinks University The repo contains R scripts capable of downloading the required data files, performing the desired analysis and testing the results against a predefined value. The final objective of the assignment is to:

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

Here is a description table of the main files in this repository from which you can reproduce all others.

	File	Description
1	assignment1.Rproj	RStudio project file
2	CodeBook.Rmd	The source that generates the CodeBook.md file
3	get_data.R	R script that downloads the necessary data files
4	Makefile	Defines the several tasks that can be executed to reproduce the analysis
5	README.Rmd	The source that generates this README.md file
6	run_analysis.R	The actual analysis script
7	test_result.R	A script that tests the output file from the analysis against an expected value
8	variables.txt	A file listing all the variables present in the resulting analysis output

Reproducing the analysis

I have made an attempt to make it very easy to reproduce the analysis of this repo. If you have make at your choice of Operating system you should be able to run the make tasks to get everything done for you with very little effort. RStudio is capable of running make tasks directly from the GUI under the Build tab (close to the Environment tab where all the variables are shown).

The Makefile in this repo has tasks to download the data files from the internet, perform the analysis and test it against a predefined result. It also contains tasks to clean up your local clone of the repo, update the documentation, including the one you are reading and the CodeBook.md file, which contains the definitions of the many variables that are part of this analysis.

To run make on the command line, make sure you are in the correct directory and then you can invoke the command along side the task name you want to execute, like so:

$ make depedencies # install the dependant packages: reshape2, knitr and digest
$ make all # executes each script in turn, `get_data.R`, `run_analysis.R`

If you are using make from RStudio, it is even easier: just click on the Build All button on the Build tab. Similarly, you may clean your local repo, deleting the raw data, the result data and the generated documentation by, at the same tab in RStudio, clicking on More -> Clean All. Its very common practice to Clean and Rebuild, so there is also a choice for that in the RStudio.

After make is done you should be able to see the results as a text file in the working directory with a filename of tidydata.txt

Please, note that the repository does not contain the raw data, but it does contain a script that downloads that for you. Also worth noting is that some files are generated dynamically, this includes the README.md you are reading and CodeBook.md.

You can also run tasks individually, such as make download or make analysis, which are pretty self explanatory.

Note that the downloaded zip file will be stored in a data subdirectory in the current project directory. The directory will be created if it does not exist. The expanded data files are going to be unzip in the current working directory, this behaviour can be changed by defining the extractDir variable in the get_data.R and run_analysis.R files. The download is around 60MB so depending on your internet connection speed, you may need to be patient with it.

Running it manually

In case you do not want or can't use make, here is how you can run everything manually. Note that you will have to install at least the reshape2 package, you can do that whithin R with install.packages("reshape2")

1. Open an R shell or RStudio
2. Make sure your current working directory is this repo
3. Source the get_data.R file to download and unzip the required data files

• source("get_data.R")

4. Source the run_analysis.R file to run the analysis. The result file named tidydata.txt should be created in the working directory

• source("run_analysis.R")

5. Optionally compare the content of the generated tidydata.txt with the expected content by sourcing test_result.R

• source("test_result.R")

source("get_data.R")
source("run_analysis.R")
source("test_result.R")

## [1] "Test PASSED 8161c06dbe59f0131a5840917d275feb"