Intro to Data Science

Lecture 6 – Visualization I

A Guide to Your Process

Scheduling

Learning Objectives

Practice

Supporting Information

Class Discussion

Today’s Plan

• Function Tutorial Debrief
• Data Visualization with ggplot2
  • Core ggplot
  • Adding geometries
  • Multiple geometries
  • Setting color
  • Customizing colors

Today’s Learning Objectives

After today’s session you will be able to:

• Discuss presentations and articulate plans for revision
• Create ggplot2 graphs
• Modify ggplot2 graph aesthetics and customize labels / colors

Function Tutorial Debrief

• How did y’all feel that went?

• What do you plan on doing differently for the 2^nd presentation?

• What questions do you have about the revision process / 2^nd draft?

Data Visualization

• Fundamental part of scientific process

• Important for:
  • Figures in papers / presentations
  • “Eyeball test” of statistical results
  • Identifying errors in data (e.g., unreasonably high/low points, typos, etc.)

• Note on word choice
  • Visualization == figures == graphs == plots
  • “Figures” are implicitly publication-quality but fundamentally still graphs

Data Viz in R

Two main options for data viz in R:

Base R

• From base R
• Simple but functional
• Base R function syntax

ggplot2

• From ggplot2 package
• Modular functions allow range of complexity
• Syntax similar to tidyverse but not identical
• Name derived from Grammar of Graphics

Data Viz in R

Two main options for data viz in R:

Base R

ggplot2

Plot Structure: ggplot2

• Requires three components (+ optional fourth)

1. Data object to plot

2. Mapping aesthetics
   • E.g., which column is on each axis, etc.
   • I.e., which variable is “mapped to” a given plot component

3. One or more geometries
   • Determines what type of plot you have

4. Theme elements
   • Controls plot-level formatting

Core ggplot Creation

• Core plot is just data object + aesthetics
  • Tells ggplot to create a plot with specified axes

• Data object is inherited by every other layer of the plot
  • So only needs to be specified once!

• What aesthetics can you specify?
  • X/Y axes
  • Color(s) of geometries

Core Graph Syntax

• Fundamental graph syntax requires two functions:
  1. ggplot
  2. aes

• Check out this example:

# Make a simple `ggplot2` plot
ggplot(data = my_df, mapping = aes(x = x_var, y = y_var))

Core ggplot2

• Get prepared for this practice
  • Create a script for this week
  • Download and read in the “minnow.csv” data

• Using ggplot2, make a graph with the minnow data where:
  • Fish species is on the X axis
  • Diameter of fish nest is on the Y axis

• What does the resulting graph look like?

Core ggplot2

Plot Type & Geometries

• Why does the plot not have anything on it?
  • Because ggplot2 needs you to specify your geometry!

• Geometries are functions you add to a plot to make the desired plot type
  • All start with geom_...
  • E.g., geom_bar, geom_point, etc.

• Geometry determines the type of plot
  • E.g., bar plot, scatterplot, etc.

Adding Elements

• Use + to add geometries to a plot

• Example syntax:

# Make a simple `ggplot2` plot
ggplot(data = my_df, mapping = aes(x = x_var, y = y_var)) +
    # Make it a scatterplot
    geom_point()

• This syntax is unique to ggplot2
  • Refers to stacked layers of plot information

Geometries

• Let’s practice adding geometries!
  • Copy the code you wrote for the previous graph
  • Add a + to the end of the line
  • In the next line add geom_point()

• What does that give you?

• Copy that code and change geom_point() to geom_boxplot()
  • What do you have now?

Geometries

Temperature Check

How are you Feeling?

Geometries Cont.

• Geometries “know” what data to use because of your core plot
  • I.e., in your top-level ggplot and aes functions

• Geometries do support arguments but minimal graphs don’t use them

• Mappings/aesthetics inherited from top to bottom

Multiple Geometries

• You can add multiple geometries to the same plot!
  • But order matters!

• Geometries added later are “in front” of earlier geometries

• Similar to how first geometry is “in front” of core ggplot

Multiple Geometries

• Make a graph with both geom_boxplot and geom_point
  • Add a + after whichever you put first, then put the other

• What happens if geom_boxplot is first?

• Versus if geom_point is first?

Multiple Geometries

See how points are “behind” boxplots on the left?

Axis Titles

• Axis titles default to column name passed to aes

• Good column names are usually not good plot axis labels!

• Column names should have no spaces / may or may not be capitalized
  • Plot axes should have spaces and be at least somewhat capitalized
  • Units may be in parentheses

Manual Axis Labels

• Can set labels manually to be prettier with labs function!

• labs has arguments x and y that expect characters to put as titles

• Example syntax:

# Make a simple `ggplot2` plot
ggplot(data = my_df, mapping = aes(x = x_var, y = y_var)) +
    # Make it a scatterplot
    geom_point() +
    # Add custom axis labels
    labs(x = "Custom X Label", y = "Custom Y Label")

Axis Labels

• Copy your code for the plot with:
  • Both a boxplot and points
  • Points in front of boxplots

• Use labs to do do the following:
  • Capitalize “species” and “diameter”
  • Put “cm” in parentheses on the y-axis

• What does that graph look like?

Axis Labels

Manual Label Cautionary Note

• If you mis-apply the labels your plot will still work but will be wrong

• Same plot but flipped labels and no way to know which is correct!

Coloring Geometries

• You can color geometries by other columns in the data!
  • You just need to pass them to the color or fill aesthetics

• Example syntax:

# Make a plot where the color and y-axis are mapped to the same variable
ggplot(data = my_df, mapping = aes(x = x_var, y = y_var, color = y_var)) +
    # Make it a scatterplot
    geom_point() +
    # Add custom axis labels
    labs(x = "Custom X Label", y = "Custom Y Label")

• Color != Fill
  • Color = borders / solid points
  • Fill = interior of shapes / points

Geometry Color

• Take the plot you created during the previous practice:
  • What happens if you map color to species in the aes call at the top?

• Change color to fill. Now what does the plot look like?

Geometry Color

color = species

fill = species

Geometry Color

• What happens if you map species to both color and fill?

• Try it and find out!

Geometry Color

Customizing Colors

Finding Fun Colors

• Coolors (coolors.co/palettes/popular)
  • Many options + search bar to narrow results

• Color Brewer 2.0 (colorbrewer2.org)
  • Fewer options but checkbox for colorblind safe palettes only

• Colors identified as hexadecimal codes
  • Hexadecimal structure: #RRGGBB

Hexadecimal Aside

• Hexadecimal = 16 digits
  • 0-9 + a-f

• Red/Green/Blue hues can be between 0 and 255
  • “Colors” are combinations of 0-255 of R/G/B

• Color with regular numbers = #RRRGGGBBB
  • If using hexadecimals: three fewer numbers to store in a computer
  • Each color x 10³-10⁶s of pixels would compound “extra” digits’ memory demands

Manually Setting Colors

• Use scale_fill_manual() or scale_color_manual()
  • Each has one argument: values

• Needs a named vector of hexadecimal codes

c("name 1" = "entry 1", "name 2" = "entry 2", "name 3" = "entry 3")

   name 1    name 2    name 3 
"entry 1" "entry 2" "entry 3" 

• Example syntax:

# Make a plot where the color and y-axis are mapped to the same variable
ggplot(data = my_df, mapping = aes(x = x_var, y = y_var, color = y_var)) +
    # Make it a scatterplot
    geom_point() +
    # Add custom axis labels
    labs(x = "Custom X Label", y = "Custom Y Label") +
    # Customize colors
    scale_color_manual(values = c("name 1" = "#00FF00", "name 2" = "#FF0000", "name 3" = "#0000FF"))

Set Colors

• To the graph you made in the previous practice:
  • Make species fill with custom colors

• Visit one of the sites I suggested to pick your colors
  • Coolors (coolors.co/palettes/popular)
  • Color Brewer 2.0 (colorbrewer2.org)

• What does that final plot look like?

Set Colors

Temperature Check

How are you Feeling?

Upcoming Due Dates

Due before lab

(By midnight)

• Muddiest Point #6

Due before lecture

(By midnight)

• Nothing!