R Workshop dplyr() lesson

dplyr_lession.R

library(dplyr)
library(tidyr)

shrub_data<-read.csv("~/dplyrLesson/shrub_volume_experiment.csv")

########################
#some basics
#######################
#ordering rows!
#no more using shrub_data=shrub_data[order(shrub_data$site),]
arrange(shrub_data, site, height)

#keep rows according to some value
#similar to subset() or shrub_data[shrub_data$site>1,]
filter(shrub_data, site>1)

filter(shrub_data, site>1, experiment=='C')

#select individual columns 
#just like shrub_data[,c('length','width')]
select(shrub_data, length, width)

#make new columns with calculations
#the volume of all shrubs?
mutate(shrub_data, volume=length*width*height)

#transmute does the same thing, but keeps only the new columns
transmute(shrub_data, volume=length*width*height)

#You can calculate any number of values at once
transmute(shrub_data, volume=length*width*height, cover=length*width)

#Exercises: 
#1) Get only shrub data from experiment A.

#2) Extract all shrubs with heights greater than 4. 

#######################
# Tie different operations together using pipes
######################

#Pipes take the output from one thing (either a function or dataframe) and make them the input to another

shrub_data %>% filter(site>1)

shrub_data %>% filter(site>1) %>% mutate(volumn=length*width*height)

#Make them easier to read by putting only one function per line
shrub_data %>% 
  filter(site>1) %>% 
  mutate(volume=length*width*height)


#Exercises:
#1) Calculate the volume of only shrubs with height greater than 4. 

######################
#Now for some actually usefull info. summarize data by statistics
#####################
#Calculate the volume, then get the mean and standard deviation from it. 
shrub_data %>% 
  mutate(volume=length*width*height) %>%
  summarise(volumeMean=mean(volume), volumeSD=sd(volume))


#But what about calculating statistics for each of our sites?
#Pipe it through the group_by() function, and specify the column you want to group by
shrub_data %>% 
  mutate(volume=length*width*height) %>%
  group_by(site) %>%
  summarise(volumeMean=mean(volume), volumeSD=sd(volume), n=n(), se=sd(volume)/sqrt(n()) )


#Exercise
# Calculate the mean and standard deviation of height and volume for each experiment group. 






######################
#More complex example using a large dataset
#####################
#The portal dataset is from 25 years of monthly rodent trapping on 24 different plots.
#Each row is a single rodent with information on date, plot number, species, sex, and other characteristics
rodents<-read.csv("~/dplyrLesson/PortalMammals_main.csv")

#1st a few more tools using summarise()

# n() can be used inside summarise() to count the number of samples inside a group. 
#Example. How many rodents total were caught in each plot in June, 1999
rodents %>%
  filter(yr==1999, mo==6) %>% #Select only entries from June, 1999
  group_by(plot) %>%  #Use group by to 
  summarise(n=n())

#Up until now we've had a small number of rows printed to the screen. Now that there are more than 10
#(the rodent data has 24 plots, so this output has 24 rows), dplyr is trying to be nice with the output
#and not fill up our console. Lets assign this output to a new dataframe to look at it fully.

june1999Counts<-rodents %>%
  filter(yr==1999, mo==6) %>% 
  group_by(plot) %>% 
  summarise(n=n())

#Now for the same period June 1999, I want to know how many of each species was caught. 
june1999Counts<-rodents %>%
  filter(yr==1999, mo==6) %>% 
  group_by(plot, species) %>% 
  summarise(n=n())


# n_distinct() counts the number of distinct values in group.
# How many species were caught at the site in each month in 1999?
species1999 <- rodents %>%
  filter(yr==1999) %>%
  group_by(mo) %>%
  summarize(numSpecies=n_distinct(species))


#Exercises
#1) Whats the average number of rodents caught in each plot in each year?

#2) Whats the average weight of all mice in each year? (the wgt column is the weight of each rodent in grams)

#3) Graph the avgerage mouse weight over all the years of the study


#############################
# tidyr, another useful tool
#############################
#This data is from vegetation transect point counts. Each column is a plot, and 4 transects (each from 1-250dm) were done in each plot.
#Very messy!
veg<-read.csv("~/dplyrLesson/transect_data.csv")

#For easier analysis, we want 1 point per row. As in the following column setup:
# plot, transect, point, species
#
# the tidyr package is made for exactly this.

cleanVeg=veg %>%
  gather(Plot, SpeciesID, -Transect, -Point)

#Here, Plot is a new column that will take on values of the old column names. SpeciesID is a new column that will take on
#values of all the cells. -Transect and -Point tell gather() to keep those columns the same. 

#tidyr can also be used with dplyr
#Here is the most common speciesID on each plot.
mostCommon=veg %>%
  gather(Plot, SpeciesID, -Transect, -Point) %>% 
  group_by(Plot, SpeciesID) %>%
  summarise(n=n()) %>%
  top_n(1, n)

#This uses the top_n() function. which orders each group and gives the highest values of each, in this case just the highest values.

shrub_volume.csv