compute_batch

Compute a region-wide geofence crash rate across several regions in one request. It compares regions side by side without a separate compute call, and a full per-cell download, for each one.

HumanBaselines.compute_batch(
    regions: list[str],
    selections: GeofenceSelections | dict | None = None,
    *,
    summary_only: bool = True,
    **filters,
) -> BatchComputeResult

The same bound and per-call filters are applied to every region in regions. Pass filters as keyword args, a GeofenceSelections model, or a dict, under the same rules as compute.

The batch is summary-only by default, so cells is omitted. Set summary_only=False if you need each region's per-cell breakdown.

A region whose compute fails, for example on a filter value it does not support such as denominator_vmt="hpms" outside California, comes back with error set rather than failing the whole call.

Example

from humanbaselines import HumanBaselines

hb = HumanBaselines()

batch = hb.compute_batch(
    ["travis", "houston", "sf", "la"],
    outcome="police_reported",
    ego_vehicle=["cars", "light_trucks"],
    road_type=["arterial", "collector_local"],
)

for item in batch.results:
    if item.result:
        print(item.region, round(item.result.rate, 3))
    else:
        print(item.region, "→ error:", item.error)

Returns: BatchComputeResult

Field	Type	Description
results	list[BatchItemResult]	One entry per requested region, in the same order as regions.

Each BatchItemResult has:

Field	Type	Description
region	str	The region this entry is for.
result	ComputeResult | None	The rate result, or None if this region errored. cells is empty unless summary_only=False.
error	str | None	Error message when result is None, otherwise None.

See Methodology for how each rate is derived, and Errors for failure handling. Transport-level errors still raise, while per-region failures are reported inline via error.