Direct measure functions¶

Each module under zadu.measures exposes a public measure() function. Direct calls are useful for one-off research code and detailed APIs such as Gap Index regional output. The scheduled ZADU interface is preferred when multiple metrics can share exact resources.

from zadu.measures import mean_relative_rank_error, neighborhood_hit, pearson_r

mrre = mean_relative_rank_error.measure(original, projection, k=20)
pearson = pearson_r.measure(original, projection)
hit = neighborhood_hit.measure(projection, labels, k=20)

Direct functions validate public inputs and return the same scalar score dictionaries as scheduled execution. Arguments representing injected resources—such as pair_statistics, rank_comparisons, or knn_info—are internal acceleration hooks. Application code should not construct them.

Module mapping¶

Alias	Module
`tnc`	`trustworthiness_continuity`
`mrre`	`mean_relative_rank_error`
`lcmc`	`local_continuity_meta_criteria`
`nh`	`neighborhood_hit`
`ca_tnc`	`class_aware_trustworthiness_continuity`
`l_tnc`	`label_trustworthiness_and_continuity`
`nd`	`neighbor_dissimilarity`
`dtm`	`distance_to_measure`
`kl_div`	`kl_divergence`
`dsc`	`distance_consistency`
`pr`	`pearson_r`
`srho`	`spearman_rho`
`ivm`	`internal_validation_measure`
`c_evm`	`clustering_and_external_validation_measure`
`snc`	`steadiness_cohesiveness`
`topo`	`topographic_product`
`proc`	`procrustes`
`stress`	`stress`
`sn_stress`	`scale_normalized_stress`
`nm_stress`	`non_metric_stress`
`cadi`	`class_angular_distortion_index`
`gi`	`gap_index`

Import modules explicitly in reusable code instead of using a wildcard import. Parameters and return keys are collected in the measure reference.