Skip to content

Latest commit

 

History

History
108 lines (89 loc) · 3.93 KB

CodeBook.md

File metadata and controls

108 lines (89 loc) · 3.93 KB
Raw

Code book for Coursera Getting and Cleaning Data project

Whole processing is done in run_analysis.R script. Once you clone this repo you probably will need to change working directory for the script in the line:

setwd("~/GitHub/coursera-getdata-002")

You can see how required libraries are loaded:

require(data.table)
require(plyr)

Following code will unzip raw data at first run of the script. If you will not delete these files, unzipping will be skiped during next runs.

if( !any(dir() == "UCI HAR Dataset") ){
  cat("Unzipping getdata-projectfiles-UCI HAR Dataset.zip\n")
  unzip("getdata-projectfiles-UCI HAR Dataset.zip")
}

Very straightforward reading of activities and features from text files. Names are changed to be more descriptive, by default it is V1 and V2.

cat("Reading activity_labels.txt\n")
activity_labels <- read.csv("UCI HAR Dataset/activity_labels.txt", header=FALSE, sep = "")
names(activity_labels) <- c("id", "activity")

cat("Reading features.txt\n")
features <- read.csv("UCI HAR Dataset/features.txt", header=FALSE, sep = "")
names(features) <- c("id", "name")

Next is important part: looking for column IDs where "mean" and "std" words are mentioned since we need to extract only the measurements on the mean and standard deviation for each measurement

mean_and_std_IDs <- c( grep("mean", features$name, fixed=TRUE), grep("std", features$name, fixed=TRUE))

Making variable names suitable for work in R: removing brackets, commas and repeatable underscores. For example "tBodyAcc-mean()-X" become "tBodyAcc_mean_X"

features$name <- gsub("\\W", "_", features$name)            
features$name <- gsub("___", "_", features$name)  
features$name <- gsub("__", "_", features$name)  
features$name <- gsub("_$", "", features$name)

Folders "test" and "trainig" have similar structure, so I write one function to process them processSet and helper function to cunstruct exact path to required files:

getPath <-function(f, set) { paste0("UCI HAR Dataset/", set,"/", f, "_", set,".txt")}

Function processSet <- function(set) {...} takes set name as an only argument. It's value can be "test" or "trainig"

Reading main data and leaving only "mean" and "std" values:

  X <- read.csv(getPath("X", set), header=FALSE, sep = "")
  names(X) <- features$name  
  mean_and_std <- X[mean_and_std_IDs]

Reading activity and subject IDs

  y <- read.csv(getPath("y", set), header=FALSE, sep = "", col.names="label_id")
  subject<- read.csv(getPath("subject", set), header=FALSE, sep = "", col.names="id")

Applying descriptive activity names to name the activities in the data set, also labeling trainig/test data

  out <- cbind( data.table(activity = as.factor(activity_labels$activity[y$label_id]),
                            subject = subject$id,
                            set),
                  mean_and_std)

Loading and combining traing and test data into one data set

cat("Reading train set\n")
train <- processSet("train")

cat("Reading test set\n")
test <- processSet("test")

result = rbind(train, test)
result$set <- as.factor(result$set)

Creation of a second, independent tidy data set with the average of each variable for each activity and each subject.

independent_dataset <- ddply(result, .(activity, subject), .fun = function(x){ colMeans(x[, 4:82], na.rm = TRUE)})

Functions

getPath - function to cunstruct exact path to required files processSet - loading and proper labeling of data

Variables created during script execution

activity_labels - Activity lables from raw data. features - Features from raw data independent_dataset - Dataset for project submission mean_and_std_IDs - IDs of data columns to be processed result - joined test and train data test - test in tidy form train - train in in tidy form

Files created during script execution

UCI HAR Dataset - folder with raw data independent_tidy_data_set.txt - file with data for project submission