processing.shape_check.ShapeCheckReport

processing.shape_check.ShapeCheckReport(
    n_overlap,
    corr,
    range_ratio,
    min_corr,
    min_range_ratio,
)

Immutable result of a forecast shape plausibility check.

All numeric fields reflect the overlap intersection of y and reference; n_overlap is the count of aligned index positions.

Attributes

Name	Type	Description
n_overlap	int	Number of aligned (overlapping) index positions used for the computation. When this is below the evaluable minimum the skipped property returns True and corr / range_ratio are float('nan').
corr	float	Pearson correlation between y and reference over the overlap. float('nan') when n_overlap < min_overlap or when the computation fails (e.g. zero-variance series).
range_ratio	float	(y.max() - y.min()) / (reference.max() - reference.min()) over the overlap. float('nan') when n_overlap < min_overlap or when the reference range is zero.
min_corr	float	Minimum acceptable Pearson correlation (passed through from check_forecast_shape).
min_range_ratio	float	Minimum acceptable range ratio (passed through from check_forecast_shape).

Examples

from spotforecast2_safe.processing.shape_check import ShapeCheckReport
import math

r = ShapeCheckReport(
    n_overlap=24, corr=0.85, range_ratio=0.9,
    min_corr=0.6, min_range_ratio=0.5,
)
assert r.plausible
assert not r.skipped
print("plausible:", r.plausible, "skipped:", r.skipped)

# NaN correlation -> not plausible
r_nan = ShapeCheckReport(
    n_overlap=24, corr=float("nan"), range_ratio=0.9,
    min_corr=0.6, min_range_ratio=0.5,
)
assert not r_nan.plausible
print("NaN corr -> plausible:", r_nan.plausible)

plausible: True skipped: False
NaN corr -> plausible: False