Altair Cheat Sheet Python: Data Visualization Reference
Altair Cheat Sheet Python
01 Build a chart
1. Import + general pattern
import altair as alt
chart = (
alt.Chart(data).mark_…()
.encode(x=…, y=…, color=…)
.transform_…(…).properties(…)
.interactive()
)
2. Choose a chart type
mark_bar bars / aggregated comparison
mark_line trends over an ordered axis
mark_circle scatter plots
mark_area filled trends / totals
mark_rect heat maps
mark_arc pie or donut charts
mark_boxplot distribution summaries
mark_geoshape geographic shapes
Also: mark_point(), mark_tick(), mark_text(), mark_rule(), mark_square(), mark_errorbar().
3. Field data types
Append a code to a field name. Describes meaning of data, not just type.
:Q quantitative – numeric magnitude
:N nominal – unordered category
:O ordinal – ordered category
:T temporal – date or time
:G GeoJSON geographic data
4. Encode fields
Short form is compact; channel objects add options.
.encode(
x=alt.X(“Horsepower:Q”, title=”Horsepower”),
y=”Miles_per_Gallon:Q”, color=”Origin:N”,
size=”Acceleration:Q”, shape=”Origin:N”,
opacity=alt.value(0.7),
tooltip=[“Name:N”, “Origin:N”]
)
x / y horizontal / vertical position
color fill or stroke by field
size mark area or line width
shape symbol category
opacity transparency
tooltip values shown on hover
row / column small-multiple layout
5. Axes, scales, & titles
x=alt.X(
“Horsepower:Q”,
scale=alt.Scale(domain=[0, 250]),
axis=alt.Axis(title=”Horsepower”, format=”.0f”))
# scale=alt.Scale(zero=False); axis=None hides axis
.properties(
title=alt.Title(“Horsepower vs. Fuel Efficiency”,
subtitle=”Automobile dataset”),
width=600, height=400)
Note: May need to use explicit types if Altair infers the wrong visual meaning.
02 Transform & Polish
6. Bin values & histogram
alt.Chart(cars).mark_bar().encode(
x=alt.X(“Horsepower:Q”, bin=True),
y=”count()”)
# bin=alt.Bin(maxbins=20) sets bin count
7. Aggregate
alt.Chart(cars).mark_bar().encode(
x=”Origin:N”, y=”mean(Miles_per_Gallon):Q”)
# mean, median, sum, min, max, count
# alt.Y(“Miles_per_Gallon:Q”, aggregate=”mean”)
8. Sort categories
x=alt.X(“Origin:N”, sort=”-y”)
# “ascending” / “descending”; “-x” / “-y”
# [“USA”, “Europe”, “Japan”] = custom order
9. Filter rows
.transform_filter(
(alt.datum.Horsepower > 100) &
(alt.datum.Origin == “USA”))
# | is OR; ~ is NOT
10. Calculate a new field
.transform_calculate(
Efficiency=”datum.Miles_per_Gallon / “
“datum.Horsepower”
).encode(y=”Efficiency:Q”)
ORDER MATTERS
Transforms run in sequence. Filter before aggregating if only filtered rows should contribute.
11. Common mark options
.mark_circle(size=70, opacity=0.7)
.mark_line(point=True, strokeWidth=2)
.mark_bar(cornerRadius=3)
.mark_arc(innerRadius=50) # donut
.mark_text(align=”left”, dx=4)
12. Complete scatter plot
chart = (
alt.Chart(cars)
.mark_circle(size=70, opacity=0.7)
.encode(
x=alt.X(“Horsepower:Q”, title=”Horsepower”),
y=alt.Y(“Miles_per_Gallon:Q”, title=”Miles per gallon”),
color=”Origin:N”,
tooltip=[“Name:N”, “Origin:N”,
“Horsepower:Q”, “Miles_per_Gallon:Q”])
.properties(title=”Horsepower vs. MPG”,
width=600, height=400)
.interactive()
)
13. Global styling
styled = (chart
.configure_view(stroke=None)
.configure_axis(grid=False)
.configure_legend(titleFont2)
.configure_title(font6))
03 Interact & Share
14. Zoom & pan
chart.interactive()
# Interval selection bound to x/y scales
15. Click to highlight
pick = alt.selection_point(fields=[“Origin”])
chart = (
alt.Chart(cars).mark_circle().encode(
x=”Horsepower:Q”, y=”Miles_per_Gallon:Q”,
color=alt.condition(pick, “Origin:N”,
alt.value(“lightgray”)))
.add_params(pick)
)
16. Brush one chart; filter another
brush = alt.selection_interval()
base = alt.Chart(cars)
points = base.mark_circle().encode(
x=”Horsepower:Q”, y=”Miles_per_Gallon:Q”
).add_params(brush)
bars = base.mark_bar().encode(
x=”Origin:N”, y=”count()”
).transform_filter(brush)
points & bars
SELECTION TYPES
selection_point() selects discrete marks; selection_interval() selects a rectangular
region.
17. Combine charts
a | b horizontal concatenation
a & b vertical concatenation
a + b layer marks in one view
alt.layer(a,b) explicit layering
18. Layer a regression line
points = alt.Chart(cars).mark_circle().encode(
x=”Horsepower:Q”, y=”Miles_per_Gallon:Q”)
line = points.transform_regression(
“Horsepower”, “Miles_per_Gallon”
).mark_line(color=”firebrick”)
points + line
19. Facet into small multiples
alt.Chart(cars).mark_circle().encode(
x=”Horsepower:Q”, y=”Miles_per_Gallon:Q”,
column=”Origin:N”)
# Or: chart.facet(column=”Origin:N”)
20. Save & Export
chart.save(“chart.html”) # interactive
chart.save(“chart.png”) # static
chart.save(“chart.svg”)
chart.save(“chart.pdf”)
chart.save(“chart.json”) # specification
# Image/offline: pip install “altair[save]”