Multiple plots from DataFrames

Introduction to Data Visualization with Julia

Gustavo Vieira Suñe

Data Analyst

Multiple variables in a plot

# Violin plot
@df insurance violin(
    :Sex, :Charges,
    label=false, linewidth=0,
    fillcolor=:grey40
)

# Add box plot
@df insurance boxplot!(
    :Sex, :Charges,
    label=false, alpha=0.75,
    fillcolor=:mediumorchid3,
    outliers=false,
)
ylabel!("Insurance Premium (USD)")

A box plot and a violin plot superimposed displaying the distribution of insurance charges by sex.

Categorical data and layouts

insurance DataFrame

Age	Sex	BMI	Children	Smoker	Region	Charges
19	female	27.90	0	yes	southwest	16884.90
18	male	33.77	1	no	southeast	1725.55
28	male	33.00	3	no	southeast	4449.46
...	...	...	...	...	...	...

Categorical column → visualize side-by-side plots
- @df recipe is compatible with the layout argument!

Layouts with DataFrames

@df insurance violin(
    :Sex,
    :Charges,
    group=:Region,
    linewidth=0,
    color=[:red :green :blue :purple],
    legend_position=:top,

    # Set layout
    layout=(2,2)

)
ylims!(0, 6*10^4)
ylabel!("Premium (USD)")

A two-by-two grid of box plots displaying the distribution of insurance charges by sex for each region.

Adding chains to the mix

@chain insurance begin
    # Smoker column to numeric
    transform(:Smoker
        => ByRow(x -> x == "yes" ? 100 : 0)
        => :Smoker)

    groupby([:Sex, :Children])
    combine(:Smoker => mean)

    @df bar(:Children, :Smoker_mean,
        group=:Sex, linewidth=0,
        fillcolor=[:cyan4 :chocolate2],
        # Set layout
        layout=2)

end
ylims!(0, 35)
xlabel!("Children")
ylabel!("Percentage of Smokers")

A one-by-two grid of bar charts displaying the percentage of smokers versus number of children.

Correlation matrix plots

A two-by-two grid of plots. The plots in the diagonal show histograms with the distributions of age and body mass index. The plot above the main diagonal sows a two-dimensional histogram of age versus BMI and the plot below the diagonal, a scatter plot of the same variables.

Correlation matrix plots

Diagonals
- Histograms of variable distributions
Above diagonal
- Two-dimensional histograms
Below diagonal
- Scatter plots with regression lines

Correlation matrix plots in StatsPlots.jl

# Using DataFrames recipe
@df insurance corrplot(

    # Numerical columns
    [:Age :BMI],

    # Customize
    markercolor=:thermal,
    fillcolor=:acton
)

¹ https://docs.juliaplots.org/latest/generated/colorschemes/

Correlation matrix plots in StatsPlots.jl

# Using DataFrames recipe
@df insurance corrplot(
    # Numerical columns
    [:Age :BMI :Children :Charges],

    # Customize
    markercolor=:thermal,
    fillcolor=:acton
)

A correlation matrix plot of age, BMI, number of children and insurance charges.

Let's practice!

Introduction to Data Visualization with Julia