Skip to content

Latest commit

 

History

History
141 lines (107 loc) · 3.27 KB

self_analysis.md

File metadata and controls

141 lines (107 loc) · 3.27 KB
Raw

Self Analysis

Mikey Saugstad
November 7, 2016

Reading Data

#
# READING SLEEP DATA
#

sleep.data <- read.csv("data/sleep_data-subset.csv")


#
# READING SELF-RECORDED DATA
#

# If having trouble with loading xlsx or rJava, try running rstudio as root
library(xlsx)

# The read.xlsx function has not been very good at coercing the data when the
# correct types are passed in (most of them are booleans), so I coerce almost
# everything to character, and then do some extra work when cleaning to get
# it all to the correct types.
classes <- c("Date", replicate(47, "character"))
self.data <- read.csv(file = "data/self_data-subset.csv", colClasses = classes)

Cleaning Data

#
# CLEANING SLEEP DATA
#

# Convert to correct data types
sleep.data$End <- as.Date(sleep.data$End, format = "%F %T")
sleep.data$Sleep.quality <- as.numeric(sub("%","",sleep.data$Sleep.quality))
sleep.data$Time.in.bed <- as.difftime(as.character(sleep.data$Time.in.bed), format = "%R")

# Use the date when I woke up to figure out which night I went to bed; I may
# go to sleep after midnight, so start date may not always give the correct
# date, but I always wake up past midnight.
sleep.data$Date <- sleep.data$End - 1


#
# CLEANING SELF-RECORDED DATA
#

# Optionally combine the drank coffee and took caffeine supplement
#mydata$coffee <- as.logical(as.numeric(mydata$coffee)) | as.logical(as.numeric(mydata$caffeine))

# Coerce remaining cols to correct types.
# TODO figured out a way to do this using col names instead of indices
binary_cols <- c(3:14,16:20,22:28,30:39,41:48)
numeric_cols <- c(2,15,29)
factor_cols <- c(21,40)
for (i in binary_cols) {self.data[,i] <- as.factor(as.logical(as.numeric(self.data[,i])))}
for (i in numeric_cols) {self.data[,i] <- as.numeric(self.data[,i])}
for (i in factor_cols) {self.data[,i] <- as.factor(self.data[,i])}

Merging Data

#
# MERGING DATASETS
#

my.data <- merge(self.data, sleep.data, by.x = "Date", by.y = "Date", all = TRUE)

Analysis

library(ggplot2)
sleep.v.sleep.qual.plot =
  ggplot(data = my.data, aes(Time.in.bed, Sleep.quality, group = sck)) +
  scale_x_continuous() +
  geom_smooth(aes(colour = sck), method = 'loess') +
  theme_bw() +
  theme(text = element_text(size = 16)) +
  labs(title = "Effect of Hours Slept on Sleep Quality",
       x = "Time in Bed (Hours)", y = "Sleep Quality (Percentage)")
sleep.v.sleep.qual.plot

summary(my.data$sck)
## FALSE  TRUE  NA's 
##   573    26     1

library(dplyr)
## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

prod.v.meditation.plot =
  ggplot(data = my.data %>% filter(!is.na(stres) & !is.na(proR)),
         aes(as.ordered(proR), group = stres)) +
  geom_bar() +
  facet_grid(droplevels(stres) ~ ., scales = 'free') +
  theme_bw() +
  theme(panel.grid.major = element_blank()) +
  theme(text = element_text(size = 16)) +
  labs(title = "Effect of Stress on Productivity",
       x = "Productivity Rating", y = "Had a Stressful Day")
prod.v.meditation.plot