tidydataproject

getting-and-cleaning-data-course-project

Creating the tidy dataset with run_analysis.R

Installation

1. Required R packages: dplyr, data.table
2. Downlaod and extract to a local folder of your choice the raw dataset from this address https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip
3. Downald the run_analysis.R script to a local folder of your choice.

Example:

$ setwd("workfolder")

-- download FUCI HAR Dataset.zip

-- unzip "FUCI HAR Dataset.zip"

-- list.dir() will return the following folders and subfolders of the "workfolder" folder:

$ list.dirs()
 [1] "."                                        "./UCI HAR Dataset"                       
 [3] "./UCI HAR Dataset/test"                   "./UCI HAR Dataset/test/Inertial Signals" 
 [5] "./UCI HAR Dataset/train"                  "./UCI HAR Dataset/train/Inertial Signals"
$ setwd("FUCI HAR Dataset")

-- list.files() will return following files

$ list.files()
[1] "activity_labels.txt" "features.txt"        "features_info.txt"   "README.txt"         
[5] "test"                "train"                                                          

Using run_analysis.R

Description

run_analysis takes as input the folder location where the raw dataset top folder is located and returns a data.frame with the tidy dataset.

Usage

run_analsys(path = ".")

Arguments

path : path (can be relative to the location of the run_analysis.R file) to the "UCI HAR Dataset" folder

Example:

Continuiong the example from above, assuming run_analysis.R has been save in the "workfolder" direcotry

$ tidydataset <- run_analysis()

will return the tidy dataset into the tidydataset data.frame

Notes

Two files will be created , one binary and the other in text csv format,

named respectivley averagedDataSet.rda and averagedDataSet.csv

Data cleaining Process - Step by Step with code

0. Setup folder names and load packages

   run_analysis(path = ".')

   # See Readme.md for usage instructions
      trainfolder <- file.path(path, "UCI HAR Dataset/train")
      testfolder  <- file.path(path, "UCI HAR Dataset/test")
      featurefile <- file.path(path, "UCI HAR Dataset/features.txt")
      activityfile <-file.path(path, "UCI HAR Dataset/activity_labels.txt")
      averagedfile <-file.path(path, "UCI HAR Dataset/averagedDataSet.")

    # R packages requirements
  library(data.table)
  library(dplyr)

1. Load Files from raw dataset

    #Train Data
    traindata <- fread(file=file.path(trainfolder,"X_train.txt"), sep = " ", data.table = FALSE, header = FALSE)
    #Test Data
    testdata <- fread(file=file.path(testfolder,"X_test.txt"), sep = " ", data.table = FALSE, header = FALSE)
    #Train Activity
    trainactivity <- fread(file=file.path(trainfolder,"y_train.txt"), sep = " ", data.table = FALSE, header = FALSE)
    #Test Activty 
    testactivity <- fread(file=file.path(testfolder,"y_test.txt"), sep = " ", data.table = FALSE, header = FALSE)
    #Train Subjects
    trainsubjects <- fread(file=file.path(trainfolder,"subject_train.txt"), sep = " ", data.table = FALSE, header = FALSE)
    #Test Subject
    testsubjects <- fread(file=file.path(testfolder,"subject_test.txt"), sep = " ", data.table = FALSE, header = FALSE)

2. Merges rows of train and test data sets

    # bind subjects and ativity columns to corresponding data set
    testcomplete <- cbind(testsubjects, testactivity, testdata)
    traincomplete <- cbind(trainsubjects, trainactivity, traindata)
    # merge rows of data sets
    completedata <- rbind(testcomplete, traincomplete)
    names(completedata)[1:2] <- c("Subject_ID","Activity")
    #remove unncessary data frames from script
    rm(traindata, testdata, trainactivity, testactivity, trainsubjects, testsubjects,testcomplete, traincomplete)    

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

See CodeBook.md for details about which variable were selected

    #Load variable names from features.text file and select the requird subset of variables 
    
    #read variables name from features.txt file
    features <- read.csv(featurefile,stringsAsFactors = FALSE, sep = " ", header = FALSE)
    features <-tbl_df(features)
    # filter measurements on the mean and standard deviation - 
      #select variables with either mean() or std(0 that start with t only)
      set1 <- filter(features, grepl("^t.*(mean()|std()).*", features$V2))
      #select variables that include either Jerk or Mag and do start with t only
      set2 <- filter(features, grepl("^t.*(Jerk|Mag).*", features$V2))
      #the desired feature variables are included in the set difference between set1 and sets
      features <- setdiff(set1, set2)
    #increase column indicies by 2 to reflect the added column with subject ids and activity data
    features$V1 <- features$V1 + 2

    #select the filtered measurement column 
   completedata <- select(completedata, c(1, 2, features$V1))
   #upate variable names with descritives names
   features$V2 <- strsplit(features$V2,"-")
   variablenames <- lapply(features$V2,FUN=function(x){if(is.na(x[3])) paste(x[2], x[1],sep="") else paste(x[2], x[1], x[3],sep="")})
   variablenames <- gsub("mean[(][)]t", "Mean", variablenames)  
                    # replace "mean()" with "Mean" and drop the letter t in front just after mean()
   variablenames <- gsub("std[(][)]t", "StdDev", variablenames) 
                    # Replace "std()" with "StdDev"and drop the letter t in front just after std()

    # assign variable names to columns
    names(completedata)[3:20] <- variablenames
    # Load activity labels key -> value map
    activitylabels <- fread(activityfile, sep = " ", data.table = FALSE, header = FALSE)
    # Replace numerics with character values from activitylabels 
    completedata[["Activity"]] <- activitylabels[match(completedata[['Activity']], activitylabels[['V1']]), 'V2']

4. Create new tidy data set with averaged means and standard deviations

    #update variable names
    variablenames <- sapply(variablenames,FUN=function(x){paste("Averaged", x,sep="")})
    names(completedata)[3:20] <- variablenames
    #group by Activity and Subject ids
    averagedDataSet <- completedata %>% group_by(Activity, Subject_ID) %>% summarise_all(funs(mean))
    #save dataset in binary and text forms
    save(averagedDataSet, file = paste(averagedfile,"rda", sep=""))
    write.csv(averagedDataSet,file = paste(averagedfile, "csv",sep=""))
    averagedDataSet