Getting and Cleaning Data Course Project
run_analysis.R
Note The code blocks used in the README are for illustration only & are incomplete. Please refer to the script for complete implementation.
This script assumes that the Samsung data is in your working directory(unzipped).
Here's a step by step explanation of how the script works along with the line numbers.
Imports necessary packages(lines[2:3])
Package reshape2 need to be imported to use melt and dcast functions used
for creating tidy data at the end of the script.
Read column names for the measurement data(lines[6])
Reads the column names from features.txt to be used with read.table() while
reading measurement data.
cNames <- read.table(file = 'UCI HAR Dataset/features.txt')Read the training and test data sets(lines[8:16])
Reads the measurement, activity and subject data using read.table(). Please
note that we are using column names stored in cNames variable while reading
the measurement data.
Merging the training and test data sets(lines[20:22])
Using rbind, we are merging the corresponding training and test data sets.
Extracting the measurements based on the mean and standard
deviation(lines[25:26])
We use grep here to identify and retrieve any measurements with mean or
standard deviation. Note that we are selecting any coulmns with either lowercase
mean or std in it.
grep('mean|std', names(x = measurement), value = TRUE)Adding descriptive activity names(lines[29:34])
activity$V1[activity$V1 == 1] <- 'WALKING'The code is pretty much self explanatory to what it is doing here.
Adding descriptive variable names(lines[37:45])
Using gsub, we are modifying the variable names as per the descriptions
provided with features_info.txt.
Merging the subject and activity columns with measurement
data(lines[50:52])
mergedData <- cbind(meanStdColumns, activity, subject)Uses cbind. Code is self explanatory.
Making a tidy data set and writing to a txt file(lines[55:59])
We use the melt and dcast functions from the reshape2 package we imported
earlier to create a tidy data set with the average of each variable for each
activity and each subject. Finally we use the write.table() funtion with
row.names set to FALSE to create a tidydataset.txt.
meltedData <- melt(data = mergedData, id = c('subject', 'activity'))
tidyData <- dcast(data = meltedData, formula = subject + activity ~ variable, mean)
write.table(x = tidyData, file = 'tidydataset.txt', row.names = FALSE) Reading the tidy data set uploaded to Coursera
We can use the read.table() function in the following way to produce a tidy
data set in our enviornment
tidyData <- read.table('./TidyData/tidydataset.txt', header = TRUE)
If you use the dim() function on tidyData, you can see that it contains
180 observations of 81 variables.
CodeBook.md
CodeBook.md file in the repo contains descriptions of variables, data and any
transformations performed to clean up the data.