Figure 5_Behavior Proportions in Cells with Dave Reineke.Rmd

# This script makes a stacked boxplot showing the proportions of behaviors at any given time.
# Kathryn Busby and Dave Reineke
# 2/26/2020
# mkbusby@email.arizona.edu

# Import libraries.

library(tidyverse)
library(ggplot2)
library(ggthemes)
library(extrafont)
library(plyr)
library(scales)
library(reshape2)
library(lubridate)

# Import data file.

data <- read.csv(file="[Insert your path to Census.csv here]",
                 header = TRUE,
                 sep = ",")

# To reorder the plot so that it uses whole datetimes instead of just hours and minutes,
# we need to create a date-time object and use it to sort.
# The Lubridate package is tidyverse adjacent and accomplishes this.

data$time2 <- data$time

data$time <- hm(data$time2)

# drop the seconds
data$time <- str_remove_all(data$time, " 0S")

# manually add in 0H where missing
data[93:96,"time"] <- "00H 10M"
data[97:100,"time"] <- "00H 23M"
data[169:172,"time"] <- "00H 19M"
data[173:176,"time"] <- "00H 52M"

# manually add zeros where needed
data[9:12,"time"] <- "14H 01M"
data[21:24,"time"] <- "15H 09M"
data[29:32,"time"] <- "16H 04M"
data[45:48,"time"] <- "17H 02M"
data[61:64,"time"] <- "20H 05M"
data[101:104,"time"] <- "01H 57M"
data[105:108,"time"] <- "02H 05M"
data[109:112,"time"] <- "02H 07M"
data[113:116,"time"] <- "04H 15M"
data[117:120,"time"] <- "04H 31M"
data[121:124,"time"] <- "05H 02M"
data[133:136,"time"] <- "19H 04M"
data[177:180,"time"] <- "01H 26M"
data[185:188,"time"] <- "11H 08M"

t.data <- data %>%
  mutate(datetime=make_datetime(year, month, date))

t.data$datetime <- paste(data$date, data$time)

# order times
t.data$order.var  <- factor(seq(1:204), labels=t.data$time2)

# Use melt to get data into long format for stacking behavior categories.

melt.data <- t.data[,-c(6:8,14)]
melt.data <- melt(melt.data, id=c("year", "month", "date", "day.night", "time", "datetime","order.var"))
melt.data$datetime <- as.factor(melt.data$datetime)

# Lay down the bones of the plot:

barplot <- ggplot() + geom_bar(aes(y = melt.data$value,
                                   x = melt.data$order.var, fill=melt.data$variable),
                               data = melt.data,
                               stat="identity")
  # This would put a line of the sample sizes across the top, but I got rid of that.
  #geom_text(aes(t.data$datetime, y=17.4, label = data$total), position = position_dodge(.9), size=3)

# Customize colors: c("black", "cadetblue3", "darkgoldenrod1", "firebrick1")
# Customize labels and title
# Customize theme (to get rid of default gray grid)

barplot + scale_fill_manual(name="Behavior", values=c("black", "white","cadetblue3", "darkgoldenrod1", "firebrick1"),
                            labels=c("abdomen dv", " ", "cleaner", "feeder", "heater")) +
  labs(x="Timestamp", y="Number of Occurrences of Behavior") +
  ggtitle("Behaviors Occurring Across Time") +
  scale_y_continuous(breaks=c(1:17)) +
  theme(plot.margin=unit(c(0.5,0,0.5,1),"cm"), panel.background = element_blank(),
        plot.title = element_text(vjust=2, size=16),
        axis.text.x = element_text(angle = 45, hjust = 1))

# Save the output plot!

ggsave(filename="[Insert path to output file here]", height=9, width=12, units="in")