Your First Application

Development approach

Start static, then add interactivity.

1. Write the visualization you want
2. Identify what the user should be able to change
3. Wrap inputs around those variables
4. Connect inputs to outputs

Step 1 — static plot

from palmerpenguins import load_penguins
from plotnine import aes, geom_point, ggplot, theme_minimal

dat = load_penguins().dropna()

(
    ggplot(dat, aes(x="bill_depth_mm", y="body_mass_g", color="species"))
    + geom_point(alpha=0.7)
    + theme_minimal()
)

Step 1b - parameterize your output

from palmerpenguins import load_penguins
from plotnine import aes, geom_point, ggplot, theme_minimal

dat = load_penguins().dropna()

x = "bill_depth_mm"
y = "body_mass_g"

(
    ggplot(dat, aes(x=x, y=y, color="species"))
    + geom_point(alpha=0.7)
    + theme_minimal()
)

Step 2 — add Shiny imports

from palmerpenguins import load_penguins
from plotnine import aes, geom_point, ggplot, theme_minimal
1from shiny.express import input, render, ui

dat = load_penguins().dropna()

1

Import input, render, and ui from shiny.express

Step 3 — discover the axis choices

from palmerpenguins import load_penguins
from plotnine import aes, geom_point, ggplot, theme_minimal
from shiny.express import input, render, ui

dat = load_penguins().dropna()
1num_cols = dat.select_dtypes("float64").columns.tolist()

1

Automatically finds all numeric columns — no hard-coded list needed

You can still use your normal Python skills!

Step 4 — add inputs

#| '!! shinylive warning !!': |
#|   shinylive does not work in self-contained HTML documents.
#|   Please set `embed-resources: false` in your metadata.
#| standalone: true
#| components: [editor, viewer]
from palmerpenguins import load_penguins
from plotnine import aes, geom_point, ggplot, theme_minimal
from shiny.express import input, render, ui

dat = load_penguins().dropna()
num_cols = dat.select_dtypes("float64").columns.tolist()

ui.input_select("x", "", num_cols, selected="bill_depth_mm")
ui.input_select("y", "", num_cols, selected="body_mass_g")

1. Dropdown for the X axis — input ID is "x"
2. Dropdown for the Y axis — input ID is "y"

Step 5 — make the output reactive

#| '!! shinylive warning !!': |
#|   shinylive does not work in self-contained HTML documents.
#|   Please set `embed-resources: false` in your metadata.
#| standalone: true
#| components: [editor, viewer]
from palmerpenguins import load_penguins
from plotnine import aes, geom_point, ggplot, theme_minimal
from shiny.express import input, render, ui

dat = load_penguins().dropna()
num_cols = dat.select_dtypes("float64").columns.tolist()

ui.input_select("x", "", num_cols, selected="bill_depth_mm")
ui.input_select("y", "", num_cols, selected="body_mass_g")


@render.plot
def plot():
    return (
        ggplot(dat, aes(x=input.x(), y=input.y(), color="species"))
        + geom_point(alpha=0.7)
        + theme_minimal()
    )

1. @render.plot tells Shiny this function produces a plot
2. input.x() and input.y() read the current dropdown selections

What just happened?

• input.x() and input.y() are reactive values — they change when the user picks
  • input.*() - signals reading from an input component
• @render.plot creates a reactive context — it re-runs whenever its inputs change
  • @render* - signals an output component
• Both dropdowns drive the same output with no extra wiring

Shiny tracked both dependencies automatically.

Important

Reactive behaviors must stay within a reactive context.

• We will talk about @reactive.*() values for reactive intermediates

Output without reactive context

#| '!! shinylive warning !!': |
#|   shinylive does not work in self-contained HTML documents.
#|   Please set `embed-resources: false` in your metadata.
#| standalone: true
#| components: [editor, viewer]
from palmerpenguins import load_penguins
from plotnine import aes, geom_point, ggplot, theme_minimal
from shiny.express import input, render, ui

dat = load_penguins().dropna()
num_cols = dat.select_dtypes("float64").columns.tolist()

ui.input_select("x", "", num_cols, selected="bill_depth_mm")
ui.input_select("y", "", num_cols, selected="body_mass_g")


(
    ggplot(dat, aes(x=input.x(), y=input.y(), color="species"))
    + geom_point(alpha=0.7)
    + theme_minimal()
)

The complete app

#| '!! shinylive warning !!': |
#|   shinylive does not work in self-contained HTML documents.
#|   Please set `embed-resources: false` in your metadata.
#| standalone: true
#| components: [editor, viewer]

from palmerpenguins import load_penguins
from plotnine import aes, geom_point, ggplot, theme_minimal
from shiny.express import input, render, ui

dat = load_penguins().dropna()
num_cols = dat.select_dtypes("float64").columns.tolist()

ui.input_select("x", "", num_cols, selected="bill_depth_mm")
ui.input_select("y", "", num_cols, selected="body_mass_g")


@render.plot
def plot():
    return (ggplot(dat, aes(x=input.x(), y=input.y(), color="species")) +
        geom_point(alpha=0.7) +
        theme_minimal()
    )