Getting and Cleaning Data with R

This project is part of Coursera student assignment.

It consists in getting a large dataset concerning 'Human Activity Recognition Using Smartphones', cleaning, and transforming it.

Directory organization

The project is organized as follows:

• A run_analysis.R script at project root
• Project documentation: README.md, CodeBook.MD
• A ./scripts directory containing scripts performing each step of data transformation
• (A ./data script will be generated when running run_amalyis.R, hosting downloaded source data)
• A tidydata.txt file, result of analysis, this file can be generated by running run_analysis.R

Directory structure:

CodeBook.md
./data
    ...
README.md
run_analysis.R
./scripts
    extract.R
    generate-tidydata.R
    getdata.R
    labelize.R
    merge.R
    normalize.R
tidydata.txt

Main script

To Run project data analysis, please run the following script in R or RStudio

# Running script from R rather than RStudio might speed up the read.fwf() step...
# setwd('PROJECT_RETRIEVED_FROM_GIT_PATH')
source('run_analysis.R')

Transformation scripts

run_analysis.R will performed the full data analysis and transformation (at each step, we do display information on data), by including and executing the following script in this order:

• ./scripts/getdata.R
  • Will get data source, copy and unzip it in ./data directory
  • Generates first data frames to work on
  • This step takes several minutes
• ./scripts/merge.R
  • Merges train and test datasets, for all 3 datasets where interested in :
    • subject_train.txt and subject_train.txt : Individuals participating in tests (represented by a number)
    • y_train.txt and y_test.txt : Performed and measured activities
    • X_train.txt and X_test.txt : Measurements
• ./scripts/extract.R
  • Only retaining mean and standard deviation for each features (only concerns Measurement dataset)
• ./scripts/normalize.R
  • Create a factor for activities, to ease dataset understanding and manipulation
  • Factor levels: "WALKING", "WALKING_UPSTAIRS", "WALKING_DOWNSTAIRS", "SITTING", "STANDING", "LAYING"
• ./scripts/labelize.R
  • Labelize all 68 remaining features (subject, activity, and all 66 mean or standard deviation for each measurement)
• ./scripts/generate-tidydata.R
  • Creates a final data frame
    • grouping subject and activity
    • computing average value for each mean or standard deviation for each measurement
  • Generates final tidydata !!!
    • tidydata.txt

You're done, a tidydata.txt file has been generated and is available for further analysis

tidydata.txt