prebrov / getdata_courseproject

Coursera's Getting-and-Cleaning-Data Course Project

GetData_CourseProject

Repositoty contains peer-assessed Coursera's Getting-and-Cleaning-Data Course Project.

David's project FAQ has been used as guidelines for the project. [1]

run_analysis.R is a script that performs the following:

1. Downloads data from https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip
2. Unzips it into ./data directory
3. Each of the files for Subject, X and Y, both from test and train data sets is then read into its own data frame using read.csv
4. All data frames are merged using rbind to combine test and train datasets and cbind to combine Subject, Activity and Measurements.
5. Column are given meaningful names using hard-coded values for Subject and Activity and those read programmatically from features.txt data set. [2]
6. Using grepl finds all column names containing "mean" or "std". I am excluding columns containing "meanFreq" since functionaly it isn't a measurement of the data from the sensors, but a rate of measurement.
7. Reads IDs and names of activities from activity_labels.txt and uses factor to convert activity ids in the dataset to descriptive names of activities. Activity IDs are used as factor levels and activity names are used as labels.
8. Uses ddply to create a second data set with the average of each variable for each activity and each subject. Data is split by variables Subject and Activity, colMeans are used to calculate means on all columns except first two (Subject and Activity).
9. Uses write.csv to save the final tidy dataset as a CSV file names ./data/tidydata.txt. File can be read back with read.csv("./data/tidydata.txt")

References:

[1] https://class.coursera.org/getdata-007/forum/thread?thread_id=49

[2] https://class.coursera.org/getdata-007/forum/thread?thread_id=99#post-442