LAS6292 live-coding: data summaries and figs

qaqc3_live_coding.R

# load the libraries ------------------------------------------------------

library(tidyverse)

# Load the csv file with data ---------------------------------------------

las_data <- read_csv(" ------ .csv") # you will need to add the path
# las_data <- read_csv("./instructor-materials/class-sessions/07-QAQC3/demo_datasets/las_data_viz.csv")

# Basic Summary of the columns in the df

# Base-R
summary(las_data)

# need to convert species to a factor
# Base-R
las_data$Species <- as.factor(las_data$Species)
# Tidyverse
las_data <- las_data %>% mutate(Species = as.factor(Species))

# Summarize again...can you spot the diffference?
summary(las_data)

# can summarize individual columns
summary(las_data$Sepal.Length)

# can have it provide only the mean
mean(las_data$Sepal.Length)
# ...or median...
median(las_data$Sepal.Length)
# or even the range...
range(las_data$Sepal.Length)
# summary of factor gives the sample size of each
summary(las_data$Species)


# That said, the tidy version of getting summaries is really useful and flexible
# It is easier for more complex combinations, so we will use that most of the time
# if we are looking at multiple variables

# %>% = "then" (this is the 'pipe' operator,
# it connects multiple tidyverse commands)

# the syntax to get the mean of SepalLength is as follows:
# "with your dataframe, then summarize the mean of column Sepal.Length"

las_data %>%
  summarize(mean = mean(Sepal.Length))

# Note: "mean=" is giving this new column the name "mean". you can call it whatever you want.
# What will it be if you don't give it any name, like in the following?
las_data %>% summarize(mean(Sepal.Length))
# it still works, but has a horrible long default name...so get in the habita of naming

# you can assign the value to a new variable
mean_sl <- las_data %>% summarize(mean = mean(Sepal.Length))
mean_sl

# you can do an operation on multiple columns in the same command.
# For ex, here we are getting the mean of both Sepal.Length and Petal.Length
mean_lengths <- las_data %>%
  summarize(
    mean_sl = mean(Sepal.Length),
    mean_pl = mean(Petal.Length)
  )
mean_lengths

# But recall, there were different species. All the previous measurements
# were pooling the values for the different species to get the average.
# How do you get the mean for each individual species?

# To get the mean for a group, use the group_by command
mean_sl <- las_data %>%
  group_by(Species) %>%
  summarize(mean_sl = mean(Sepal.Length))
mean_sl

# incredibly valuable, because this is the start of a table...
# for instance, you can get the mean, sd, and range
mean_sl <- las_data %>%
  group_by(Species) %>%
  summarize(
    mean_sl = mean(Sepal.Length),
    sd_sl = sd(Sepal.Length),
    range_sl = range(Sepal.Length)
  )
mean_sl
# huh? it adds a row to each group with the min and max.
# this isn't very tidy, is there a better way to do this?

# yes! get the min and the max for each column (range is the diff of these 2)
mean_sl <- las_data %>%
  group_by(Species) %>%
  summarize(
    mean_sl = mean(Sepal.Length),
    sd_sl = sd(Sepal.Length),
    min_sl = min(Sepal.Length),
    max_sl = max(Sepal.Length)
  )
mean_sl

# BUT THERE IS AN ISSUE WITH JUST LOOKING AT SUMMARIES

# This is to do the Datasaurus comparison.
# https://cran.r-project.org/web/packages/datasauRus/vignettes/Datasaurus.html
install.packages("datasauRus")
library(datasauRus)

if (requireNamespace("dplyr")) {
  suppressPackageStartupMessages(library(dplyr))
  datasaurus_dozen %>%
    group_by(dataset) %>%
    summarize(
      mean_x    = mean(x),
      mean_y    = mean(y),
      std_dev_x = sd(x),
      std_dev_y = sd(y),
      corr_x_y  = cor(x, y)
    )
}

if (requireNamespace("ggplot2")) {
  ggplot(datasaurus_dozen, aes(x = x, y = y, colour = dataset)) +
    geom_point() +
    theme_classic() +
    theme(legend.position = "none") +
    facet_wrap(~dataset, ncol = 3)
}




# ALWAYS PLOT THE DATA!!!!

# histograms in base R

hist(las_data$Sepal.Length)
hist(las_data$Sepal.Width)
hist(las_data$Petal.Length)

# Box plots in base r
boxplot(las_data$Sepal.Length)
boxplot(las_data$Petal.Length)

# ALL OF THIS IS GREAT, BUT GRAPHS CAN OBSCURE PATTERNS TOO
# ALWAYS PLOT THE RAW DATA
# with correlated data, can do this really well
# VIEW A on Y, and THAT IS INCREDIBLY USEFUL

# scatterpolot in base r:  plot(x,y)
plot(las_data$Sepal.Width, las_data$Sepal.Length)


# BUT WE HAVE BEEN LOOKING THE WHOLE TIME WITH ALL SPECIES TOGETHER. LETS
# LOOK AT THE PATTERN FOR ALL SPECIES SEPERATELY.
# This is much easier to do in ggplot2 (tidyverse)-

# Here is how to do histograms in ggplot2
# https://r-graph-gallery.com/220-basic-ggplot2-histogram.html
plot <- ggplot(
  las_data,
  aes(x = Petal.Length)
) +
  geom_histogram()
plot

# easy to customize. Change the background with different "theme"
plot <- ggplot(
  las_data,
  aes(x = Petal.Length)
) +
  geom_histogram() +
  theme_classic()
plot

# can change colors
plot <- ggplot(
  las_data,
  aes(x = Petal.Length)
) +
  geom_histogram(
    binwidth = 0.5,
    fill = "red",
    color = "black"
  ) +
  theme_classic()
plot


# different plot for different groups
plot <- ggplot(
  las_data,
  aes(
    x = Petal.Length,
    fill = Species
  )
) +
  geom_histogram() +
  theme_classic()
plot

# BOOM. Finally see why that weird value set at end


# or give them each their own panel
plot <- ggplot(
  las_data,
  aes(
    x = Petal.Length,
    fill = Species
  )
) +
  geom_histogram() +
  facet_grid(rows = vars(Species)) +
  theme_classic()
plot




# box plots in tidy
# https://r-graph-gallery.com/boxplot.html

las_data %>%
  ggplot(
    aes(x = Species, y = Sepal.Length, fill = Species)
  ) +
  geom_boxplot() +
  theme_bw()


las_data %>%
  ggplot(aes(x = Species, y = Sepal.Length, fill = Species)) +
  geom_boxplot() +
  geom_jitter(color = "black", size = 0.4, alpha = 0.9) +
  theme_bw()


# scatter plots (x,y) in ggplot2
# can also look at the x y values
las_data %>%
  ggplot(
    aes(x = Sepal.Width, y = Sepal.Length)
  ) +
  geom_point() +
  theme_bw()

# different color points for each species

las_data %>%
  ggplot(
    aes(x = Sepal.Width, y = Sepal.Length, color = Species)
  ) +
  geom_point() +
  theme_bw()

# if you want to remove the "error", uncomment the code below and run it.
# las_data<-las_data %>%
#   filter(Sepal.Length<12)
# you have filtered the data to ONLY allow values of Sepal.Length less than 12



# bigfoot data ------------------------------------------------------------



bigfoot <- read_csv("./instructor-materials/class-sessions/07-QAQC3/demo_datasets/annual_bigfoot.csv")



# complete code for plot with bigfoot data --------------------------------


#  make a scatter plot ----------------------------------------------------

# first convert year to an integer
bigfoot$year <- as.integer(bigfoot$year)
# MAPPING ggplot(....)+
# GEOM (geom_point....)

# if you want to add the options in the geom_point, don't foget to
# add a , (comma) after each line except the last one.
# Also, don't forget the last )

scatter <- ggplot(
  data = bigfoot,
  mapping = aes(x = year, y = sightings)
) +
  geom_point(
    color = "black", # the stroke color, the circle outline
    fill = "black", # color of the circle inner part
    shape = 20 # shape of the marker
    # alpha=1, # circle transparency, [0->1], 0 is fully transparent
    # size=0, # circle size
    # stroke = 1 # the stroke width
  )

# Coordinates and Scales
# take your plot named scatter, and add some scales to y axis
scatter <- scatter +
  scale_y_continuous(breaks = seq(0, 300, 20))

# LABELS AND GUIDES
# add labes for axes and a title
scatter <- scatter +
  labs(
    x = "year",
    y = "Sightings",
    title = "Annual Bigfoot Sightings"
  )



# if you want to add a regression line
scatter <- scatter +
  # stat_smooth(method = "lm",
  stat_smooth(
    method = "loess",
    formula = y ~ x,
    geom = "smooth",
    color = "red",
    linetype = "dashed",
    size = 1.5
  )




# THE WHOLE THING WITH ALL SECTIONS TOGETHER ------------------------------


scatter <- ggplot(
  data = bigfoot,
  mapping = aes(x = year, y = sightings)
) +
  geom_point(
    color = "black", # the stroke color, the circle outline
    fill = "black", # color of the circle inner part
    shape = 20 # shape of the marker
    # alpha=1, # circle transparency, [0->1], 0 is fully transparent
    # size=0, # circle size
    # stroke = 1 # the stroke width
  ) +
  theme_classic() +
  theme(
    plot.title = element_text(face = "bold", size = 20, family = "Arial", colour = "black"),
    # Sets title size, style, location
    # legend.position=c(0.5,0.95),
    axis.line.y = element_line(color = "black", linewidth = 0.5, lineend = "square"),
    axis.line.x = element_line(color = "black", linewidth = 0.5, lineend = "square"),
    axis.title.x = element_text(colour = "black", size = 20, vjust = -0.5), # S ets x axis title size, style, distance from axis #add , face = "bold" if you want bold
    axis.title.y = element_text(colour = "black", size = 20, vjust = 1.5), # S ets y axis title size, style, distance from axis #add , face = "bold" if you want bold
    axis.text.x = element_text(colour = "black", size = 16, angle = 45, vjust = 0, hjust = 0), # Sets size and style of labels on axes
    axis.text.y = element_text(colour = "black", size = 16, angle = 0, vjust = 0, hjust = 0), # Sets size and style of labels on axes
    # legend.title = element_blank(),                                             # Removes the Legend title
    # legend.key = element_blank(),                                              #R emoves the boxes around legend colors
    # legend.text = element_text(face="italic", color="black", size=12),
    # legend.position = "top",
    # legend.direction = 'horizontal',
    # legend.key = element_rect(colour = "black"),                              #s ets size and style of labels on axes
    # plot.margin = unit(c(1,2,1,1), "cm")
    plot.background = element_rect(fill = "white"),
    panel.background = element_rect(fill = "white")
  )

scatter

updates with edits to code and styler to format