---
title: "Boxplot, Histogram, ANOVA, ggbarplot"
author: "Scott Gabara"
date: "06/12/2020"
output:
  html_document: default
  pdf_document: default
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

## R Markdown

```{r cars}
#Base graphics boxplot example 
#I feel like weve come a long way from base graphics and just want to give an example to highlight the difference.

library(ggplot2)
boxplot(PlantGrowth[,1] ~ PlantGrowth[,2], notch = TRUE, col = c(2,3,4), xlab="Treatment", ylab="Response")
legend("topleft", y = 1,legend = c("setosa","versicolor","virginica"), col = c("red","green","blue"), pch=16)

```
You can embed an R code chunk like this:

```{r cars1}
#ggplot2 boxplot/violin plot/jitter points example
head(CO2)
str(CO2)
summary(CO2)
CO2$TypeTreatment <- paste(CO2$Type, CO2$Treatment)
ggplot(CO2, aes(x = TypeTreatment, y = uptake, fill=Type)) +
  geom_violin(alpha=0.5) +
  geom_jitter() +
  geom_boxplot(alpha=0.1) +
  stat_summary(aes(), fun.y=mean, geom="point", shape=21, size=3, fill="red",color="black", show.legend = FALSE,  position=position_dodge(.75)) +
  scale_fill_manual(values=c("Blue", "Red")) +
  scale_color_manual(values=c("Blue", "Red")) +
  xlab("Species") +
ylab(expression(~CO[2]*" uptake (umol / m"^{-2}*" sec"^{-1}*")")) +
  theme_classic()


```

```{r cars2}
#HISTOGRAM

#A histogram of uptake using ggplot2 set to density
ggplot(CO2, aes(uptake, fill = TypeTreatment)) + geom_density(alpha = 0.2) + theme_classic()

# Test for homogeneity of variance
b.test <- bartlett.test(CO2[,5] ~ CO2$TypeTreatment)
# Bartlett test for homogeneity of variances
c(b.test)
summary(b.test)
library(car) # load package to run leveneTest
leveneTest(CO2[,5] ~ CO2$TypeTreatment) # compare Sepal length variance among species

#ANOVA
model1 <- lm(uptake ~ TypeTreatment, CO2)
anova(model1)
# Bonferroni corrected T-test
bf.test <- pairwise.t.test(CO2$uptake, CO2$TypeTreatment, p.adj="bonferroni")
print(bf.test)

```

```{r cars3}
# one way ANOVA with error
library(Rmisc)
means<-summarySE(CO2, measurevar="uptake", groupvars=c("TypeTreatment"))
means
# ggplot boxplot
# Standard deviation of the mean as error bar
p <- ggplot(CO2, aes(x=TypeTreatment, y=uptake, fill=TypeTreatment)) + 
geom_boxplot() +
stat_summary(aes(), fun.y=mean, geom="point", shape=21, size=3, fill="red",color="black", show.legend = FALSE,  position=position_dodge(.75)) +
theme_classic() +
ylab(expression(~CO[2]*" uptake (umol / m"^{-2}*" sec"^{-1}*")")) +
scale_colour_brewer(palette = "Spectral") +
scale_fill_brewer(palette = "Spectral")
p
```

```{r cars4}
# Two way ANOVA boxplot with error
# this example uses the free and avaiable CO2 dataset in R
p <- ggplot(CO2, aes(x=Type, y=uptake, fill=Treatment)) + 
geom_boxplot(aes(group=TypeTreatment)) +
stat_summary(aes(group=TypeTreatment), fun.y=mean, geom="point", shape=21, size=3, fill="red",color="black", show.legend = FALSE,  position=position_dodge(.75)) +
theme_classic() +
ylab(expression(~CO[2]*" uptake (umol / m"^{-2}*" sec"^{-1}*")")) +
scale_colour_manual(values = c("Red", "Blue")) +
scale_fill_manual(values = c("Red", "Blue"))
p

```
```{r cars5}
# Two way ANOVA barplot variant with error
# this example uses the free and avaiable CO2 dataset in R
library(Rmisc)
qmeans<-summarySE(CO2, measurevar="uptake", groupvars=c("Type", "Treatment", "Plant"))
qmeans

## We need to expand Rcolorbrewer Spectral palette size was 11 but we need 12
library(RColorBrewer)
cols <- colorRampPalette(brewer.pal(12, "Spectral"))
myPal <- cols(length(unique(CO2$Plant)))

p <- ggplot(CO2, aes(x=Type, y=uptake, fill=Plant)) + 
geom_boxplot(aes(group=Plant)) +
stat_summary(aes(group=Plant), fun.y=mean, geom="point", shape=21, size=3, fill="red",color="black", show.legend = FALSE,  position=position_dodge(.75)) +
theme_classic() +
ylab(expression(~CO[2]*" uptake (umol / m"^{-2}*" sec"^{-1}*")")) +
scale_fill_manual(values = myPal) +
scale_color_manual(values = myPal)
p


```

Note that the `echo = FALSE` parameter was added to the code chunk to prevent printing of the R code that generated the plot.