Getting and Cleaning Data Project by George Drakulic

download data from

http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones

unzip it at your working directory

it will create directoy called "UCI HAR Dataset"

in that directory there are two other directories: "test" and "train",

as well as labels of variour columns of the data in txt files

create a data frame "subject" by reading subject_test.txt, which has rows of a subject's number,

and name the column "subject"

subject <- read.table("UCI HAR Dataset/test/subject_test.txt")
colnames(subject) <- c("subject")

create a data frame "activity" by reading y_test.txt, which has rows of an activity's number,

and name the column "activity"

activity <- read.table("UCI HAR Dataset/test/y_test.txt")
colnames(activity) <- c("activity")

change the numbers in activity data frame into textual activities listed in "activity_labels.txt"

activity <- factor(activity$activity, levels = c(1, 2, 3, 4, 5, 6), labels = c("walking", "walking_upstairs", "walking_downstairs", "sitting", "standing", "laying"))

merge two data frames into one (subject and activity into dat1)

dat1 <- cbind(subject, activity)

read X_test data with 561 columns

X_test <- read.table("UCI HAR Dataset/test/X_test.txt")

get the names of the features (which will be column names of X_test data)

features <- read.table("UCI HAR Dataset/features.txt")

rows should be columns, so we will transpose rows and columns, and get just the names of the features

feature_name <- t(features)[2,]

rename the columns of X_test data with the column names of the features from features.txt

colnames(X_test) <- feature_name

get the index of columns that have "mean()" in them, and also "meanFreq()

g_mean <- grep("mean()", feature_name)

get the index of columns that have "std()" in them

g_std <- grep("std()", feature_name)

combine those indices into one

g_all <- append(g_mean, g_std)

sort the indices

these are the indices we need in our data

sorted_indices <- sort(g_all)

the data set with columns of needed data of mean, and std computations

X_test_mean_and_std <- X_test[, sorted_indices]

merge subject + activity (dat1), and X_test_mean_std

X_test_merged <- cbind(dat1, X_test_mean_and_std)

just to check the final result: nrow(X_test_merged) gives me 2947,

ncol(X_test_merged) gives me 81, so the test merged data frame is 2947 x 81

Do the same for the train data

subjectTrain <- read.table("UCI HAR Dataset/train/subject_train.txt")
colnames(subjectTrain) <- c("subject")

activityTrain <- read.table("UCI HAR Dataset/train/y_train.txt")
colnames(activityTrain) <- c("activity")

activityTrain <- factor(activityTrain$activity, levels = c(1, 2, 3, 4, 5, 6), labels = c("walking", "walking_upstairs", "walking_downstairs", "sitting", "standing", "laying"))
dat2 <- cbind(subjectTrain, activityTrain)
colnames(dat2) <- c("subject", "activity")

X_train <- read.table("UCI HAR Dataset/train/X_train.txt")
colnames(X_train) <- feature_name
X_train_mean_and_std <- X_train[, sorted_indices]

X_train_merged <- cbind(dat2, X_train_mean_and_std)

just to check the final result: nrow(X_train_merged) gives me 7352,

ncol(X_train_merged) gives me 81, so the train merged data frame is 7352 x 81

merging the test and the training sets should give me (2947 + 7352) x 81 data frame

i.e. 10299 x 81

X_final <- rbind(X_test_merged, X_train_merged)

nrow(X_final) gives me 10299, which is as expected, ncol(X_final) is 81 as well

empty data frame for tidy data

X_tidy <- data.frame()

the data is split by subject, activity, and other variables, so it computes

the mean of each variable for each subject and his/her activity,

e.g. subject 1, activity 1, all means of variables; subject 2, activity 1, means,

up to last subject, activity 1

then, it starts with subject 1, activity 2 ..., last s, act 2 and so on

data_split <- split(X_final[, 3:81], list(X_final$subject, X_final$activity))
X_tidy <- sapply(data_split, colMeans)

nrow(X_tidy) = 79, ncol(X_tidy) = 180

output with row.names=F so the data is properly aligned in MS Excel, but the row names

are, of course, missing

write.table(X_tidy, file="./X_tidy.txt", sep="\t", row.names = FALSE)

output with row.names=T so the data is NOT properly aligned in MS Excel

write.table(X_tidy, file="./X_tidy_with_row_names.txt", sep="\t")

head(X_tidy, 2)