validate_model builtin

validate_model(
    model,
    grade,
    data,
    values,
    *,
    weights=None,
    domain_column=None,
    density=None,
    reference=None,
)

Statistics of a block model against the data it was estimated from, per domain and over all.

Parameters:

Name	Type	Description	Default
model	BlockModel	The blocks; weighted by their tonnage, volume × density.	required
grade	str or array_like	Column of model holding the block grades, or the grades; NaN is skipped.	required
data	(PointSet, Table or dict)	The samples (composites).	required
values	str or array_like	Column of data holding the sample grades, or the grades; NaN is skipped.	required
weights	array_like or str	Declustering weights of data; adds a declustered row that the model is compared with.	None
domain_column	str or tuple of str	Domain column of both model and data, or one name for each.	None
density	(float, array_like or str)	Block densities, or their column; default 1.	None
reference	str or array_like	Other grades on the same blocks, e.g. the truth or a previous model.	None

Returns:

Type	Description
Table	One row per domain (sorted, then "all") and source (naive and declustered data, model, reference): n, tonnage (null for data), mean, variance, cv, P10, P50, P90, mean_diff (mean / data mean - 1) and variance_ratio (variance / data variance), against the declustered data when weights are given.

used in

• units
• model checks