#Import StreamQuality_all.xlxs through 'Import Dataset'
#to view the first few lines of data to make sure that the dataset imported correctly
head(StreamQuality_all)

#add libraries
library(dplyr)
library(tidyr)
library(ggplot2)
library(scales)

#Nitrogen trend per EPA REGION? 
#First calculate the mean N of each region
stream_qualityN <- StreamQuality_all %>% 
  group_by(EPAREGION) %>%
  summarise(avgNTL = mean(NTL))   
stream_qualityN

#Next calculate the median N of each region
stream_qualityN_median <- StreamQuality_all %>% 
  group_by(EPAREGION) %>%
  summarise(medNTL = median(NTL))
stream_qualityN_median

#Create a box plot to compare across regions
p = ggplot(data = StreamQuality_all, aes(x = EPAREGION, y = NTL)) + 
    geom_boxplot()+
    theme(axis.text.x = element_text(angle = 30, hjust = 1)) + 
    labs( x="EPA Region", y="Nitrogen Total Load (ug/L)")
p

#get the quartile values R computed to create the box plot
ggplot_build(p)

#create a new box plot with a log base 10 y axis
ggplot(data = StreamQuality_all, aes(x = EPAREGION, y = NTL)) + 
  geom_boxplot()+
  scale_y_log10() +
  theme(axis.text.x = element_text(angle = 30, hjust = 1)) + 
  labs( x="EPA Region", y="Nitrogen Total Load (ug/L)")