Where deliveries end up: zone scatter and heatmaps

wa_setpieces.core.zones splits the pitch into thirds, wide/half-space/ central channels, or a configurable zone grid. Here we look at where corner and free-kick deliveries land (their end location), which is far more informative than where they start (always the corner arc / free-kick spot) – first as individual points, then as a binned grid.

from pathlib import Path

from wa_setpieces import delivery_locations, load_matches
from wa_setpieces.viz.plots import plot_zone_heatmap, plot_zone_scatter
from wa_setpieces.core.zones import add_channels

try:
    _here = Path(__file__).resolve().parent
except NameError:
    _here = Path.cwd()
DATA = _here.parent / "tests" / "data" / "sample_match.json"

events = load_matches(DATA)

Every corner’s end-location as its own point, colored by outcome and shaded by density underneath – no binning choice deciding how coarse the picture is, and success/failure is visible per point, not just an aggregate count:

corners = delivery_locations(events, "corner")
fig, ax = plot_zone_scatter(
    corners, x_col="end_x", y_col="end_y", title="Corner delivery landing spots",
)

The same idea for free kicks:

free_kicks = delivery_locations(events, "free_kick")
fig, ax = plot_zone_scatter(
    free_kicks, x_col="end_x", y_col="end_y", title="Free-kick delivery landing spots",
)

Reach for the binned grid instead when exact per-zone counts are the point, not the shape of the pattern:

fig, ax = plot_zone_heatmap(
    corners, x_col="end_x", y_col="end_y", title="Corner delivery end zones",
)

Zones aren’t just for pitch plots – add_channels works on any DataFrame with x/y columns, e.g. to see which width channel throw-ins are taken from:

throw_ins = delivery_locations(events, "throw_in")
throw_ins = add_channels(throw_ins, n=5)
throw_ins["channel"].value_counts()

channel
left_wide           28
right_wide          22
left_half_space      0
central              0
right_half_space     0
Name: count, dtype: int64