Track and Field World Record Progression

The figure below show how the track and field world records have progressed over time.

Click on the legend to highlight individual events and see the respective world record holders.

Code

import altair as alt
import polars as pl
from alltime_athletics_python.io import download_data
from alltime_athletics_python.io import import_running_only_events

# download_data()
df = import_running_only_events("./data")

Code

world_records = (
    df.filter(pl.col("event").str.contains("walk") == False)
    .filter(pl.col("event type") == "standard")
    .sort("sex", "distance", "event", "date of event")
    .with_columns(
        pl.col("result seconds")
        .cummin()
        .over("sex", "event")
        .alias("world record time")
    )
    .filter(pl.col("result seconds") == pl.col("world record time"))
    .groupby("sex", "event", "result seconds", maintain_order=True)
    .first()
    .with_columns(
        (
            100
            * pl.col("result seconds")
            / pl.col("result seconds").min().over("sex", "event")
        ).alias("percent of wr")
    )
)

world_records = pl.concat(
    [
        world_records,
        world_records.filter(pl.col("rank") == 1).with_columns(
            [
                pl.lit("2023-06-07")
                .str.strptime(pl.Date, format="%Y-%m-%d")
                .alias("date of event"),
                pl.lit(-1).cast(pl.Int64).alias("rank"),
            ]
        ),
    ]
).with_columns(pl.col("sex").apply(lambda s: s.title()))

data = world_records.select(
    "date of event", "percent of wr", "event", "sex", "name", "rank"
).to_pandas()
legend_selection = alt.selection_point(fields=["event"], bind="legend")
legend_selection_empty = alt.selection_point(
    fields=["event"], bind="legend", empty=False
)

base = (
    alt.Chart(data)
    .encode(
        x=alt.X("date of event:T")
        .scale(domain=("1950-01-01", "2026-01-01"))
        .title("Year"),
        y=alt.Y("percent of wr:Q")
        .scale(domain=(100, 110))
        .axis(values=list(range(100, 120, 2)))
        .title("Time in % of current WR"),
        color=alt.Color(
            "event:N",
            sort=world_records.sort("distance")["event"]
            .unique(maintain_order=True)
            .to_list(),
        )
        .scale(scheme="dark2")
        .title("Event"),
        # strokeDash="sex:N",
        opacity=alt.condition(legend_selection, alt.value(1), alt.value(0)),
    )
    .properties(width=800, height=500)
    .add_params(legend_selection)
    .add_params(legend_selection_empty)
)

base_no_endpoint = base.transform_filter(alt.datum["rank"] > 0)

text = base_no_endpoint.encode(
    text="name:N",
    opacity=alt.condition(legend_selection_empty, alt.value(0.9), alt.value(0.0)),
)


chart = (
    alt.layer(
        base.mark_line(interpolate="step-after", clip=True, strokeWidth=3),
        base_no_endpoint.mark_point(filled=True, clip=True, size=100),
        text.mark_text(clip=True, fontSize=14, angle=270 + 45, align="left", dx=15),
    )
    .facet(
        row=alt.Row("sex:N").title("").header(labelAngle=0),
        title="World Record Progression",
    )
    .resolve_scale(x="independent")
)

display(chart)