---
title: "Readme"
author: "Austin Overman"
date: "March 12, 2016"
output: html_document
---

### Coursera: Getting and Cleaning Data, wk 4 Assignment
Collect data, create a tidy data set, perform basic descriptive stats, and upload
to github. See assignment description for full details. Data are found here, along with description:  
https://d396qusza40orc.cloudfront.net/getdata%2Fprojectfiles%2FUCI%20HAR%20Dataset.zip
http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones
 
Use the following instructions: Please upload the tidy data set created in step 5 of the instructions as a txt file created with write.table() using row.name=FALSE.

The data was downloaded "Mon Feb 22 22:07:05 2016" and extracted without modification of the sub folders, the individual data files were accessed via calls to the sub folders they were extracted.

The script with descriptive commentary is as follows.

```{r eval = FALSE}
## setting up a variable for the original working directory to reset at the end of the script
owd <- getwd()

## setting the working directory to where the test data are saved
setwd("C:/Users/Austin/Desktop/Analytics Programs/Coursera/Cleaning Data/Course Project/UCI HAR Dataset/test")
list.files()

## loading the dply package which will be used for the majority of the tiding operations
library(dplyr)

## creating the dplyr data.frame/table_df class object for the test data set
X_test <- tbl_df(read.csv("X_test.txt", header = FALSE, sep = ""))
y_test <- tbl_df(read.csv("y_test.txt", header = FALSE, sep = ""))
sub_test <- tbl_df(read.csv("subject_test.txt", header = FALSE, sep = ""))

str(X_test)
View(X_test)

str(y_test)
View(y_test)
quantile(sub_test$V1)

## Changing the working directory to the location of the training data set
setwd("C:/Users/Austin/Desktop/Analytics Programs/Coursera/Cleaning Data/Course Project/UCI HAR Dataset/train")
      list.files()

## creating the dplyr data.frame/table_df class object for the training data set
X_train <- tbl_df(read.csv("X_train.txt", header = FALSE, sep = ""))
y_train <- tbl_df(read.csv("y_train.txt", header = FALSE, sep = ""))
sub_train <- tbl_df(read.csv("subject_train.txt", header = FALSE, sep = ""))

str(X_train)
View(X_train)
View(sub_train)
quantile(sub_train$V1)

str(y_train)
View(y_train)


## Resetting the working directory to get the variable names and activity lists 
setwd("C:/Users/Austin/Desktop/Analytics Programs/Coursera/Cleaning Data/Course Project/UCI HAR Dataset")
      list.files()

## generating vectors of the activity names and variable names to add to the acc_data table
activity <- tbl_df(read.csv("activity_labels.txt", header = FALSE, sep = ""))
variables <- tbl_df(read.csv("features.txt", header = FALSE, sep = "", 
                             stringsAsFactors = FALSE))

## renaming the column names for join functions; in activity 
activity <- rename(activity, activity = V1)
activity <- rename(activity, activity_name = V2)
activity$activity <- as.factor(activity$activity)

activity
head(variables, 10)
tail(variables, 10)

## merge the X_test and X_train data sets by row binding through dplyr. 
## x are the independant variables
x <- bind_rows(X_test, X_train)

## renaming the columns in 'x' data frame to the variables from the features.txt file
## but first the column names and set as characters class names
oldnames <- as.character(names(x))
newnames <- variables$V2
      names(x)[names(x)%in% oldnames] <- newnames

head(x[,1:10],10)

## merging y's, and s's to form 1 vector each which is then merged to the x data sets by 
## dplyr column bind to create the original full data set, 
## y is the activity list, it is the dependant varaible. s are the study subject code.
y <- bind_rows(y_test, y_train)
y <- rename(y, activity = V1)
y$activity <- as.factor(y$activity)

s <- bind_rows(sub_test, sub_train)
s <- rename(s, subject = V1)
s$subject <- as.factor(s$subject)
quantile(as.integer(s$subject))

## removing duplicate columns, subsetting the variables that contain 'mean()' and 'std()' 
acc_data <- x[ , !duplicated(colnames(x))]
acc_data <- select(acc_data, -contains("-meanFreq"))
acc_stat_m <- select(acc_data, matches(".-mean()."))
acc_stat_s <- select(acc_data, matches(".std."))

## column binding to create the reduced data frame for final analysis
acc_stat <- bind_cols(y, s, acc_stat_m, acc_stat_s)
str(acc_stat)


## adding the activity name variable and then fixing all the variable names so they
## can be renamed by the variables vector
acc_stat <- left_join(acc_stat, activity, by = "activity")
acc_stat <- acc_stat[, c(69, 1:68)]
names(acc_stat)

str(acc_stat)
View(acc_stat)
head(acc_stat[,1:5], 20)
head(acc_stat[,65:69],20)
tail(acc_stat[,65:69], 20)

## final tiding operation using melt() to transpose the signal variables to a single column
## of 'signal' for the id variable, and the value column as 'measurment'
library(reshape2)

acc_stat_long <- melt(acc_stat, variable.name = "signal", value.name = "measurement")
head(acc_stat_long)
View(acc_stat_long)
str(acc_stat_long)

## grouping the final tidy data by activity name, subject and signal and then generating
## the mean for each signal, this is the final output
avg_stat <- group_by(acc_stat_long, activity_name, subject, signal) %>% 
      summarise(average = mean(measurement))

View(avg_stat)
str(avg_stat)

## creating a txt file with write.table() using row.name=FALSE for upload to the coursera
## assignment site.
setwd("C:/Users/Austin/Desktop/Analytics Programs/Coursera/Cleaning Data/Course Project")
list.files()

write.table(avg_stat, file = "signal_avg.txt", sep = ",", row.names = FALSE)

setwd(owd)
```