Random-Exemplar Similarity

Generated by scripts/run_isolation_similarity.py.

This is an isolation-style similarity experiment, not sklearn's anomaly-detection IsolationForest. Each trial samples exemplar witnesses and feature columns, assigns every witness to the nearest sampled exemplar, and turns repeated co-assignment into a distance matrix.

Outputs

Summary

dataset trials exemplar_count feature_fraction mean_distance ultrametricity_score mae_to_cophenetic_ultrametric rmse_to_cophenetic_ultrametric spearman_vs_hamming mean_clade_similarity_separation
athanasius_epistles_data_bin_csv 500 3 0.5 0.646879 0.0136364 0.0673156 0.0952269 0.961923 0.413188
athanasius_epistles_data_multistate_csv 500 3 0.5 0.61197 0.0227273 0.0565714 0.0790267 0.939394 0.463517

Clade Separation

dataset node_id node_members node_size within_similarity cross_similarity similarity_separation support_fraction
athanasius_epistles_data_bin_csv N2 F G 2 0.908 0.2307 0.6773 1
athanasius_epistles_data_bin_csv N4 M223 M2423 2 0.862 0.2648 0.5972 1
athanasius_epistles_data_bin_csv N5 D F G 3 0.682667 0.226593 0.456074 0.833333
athanasius_epistles_data_bin_csv N3 M1739 U1 2 0.74 0.3584 0.3816 1
athanasius_epistles_data_bin_csv N1 B C 2 0.732 0.3775 0.3545 1
athanasius_epistles_data_bin_csv N6 A B C M1739 P46 U1 6 0.509733 0.278056 0.231678 0.833333
athanasius_epistles_data_bin_csv N7 A Ath B C M1739 M223 M2423 P46 U1 9 0.420556 0.226593 0.193963 0.833333
athanasius_epistles_data_multistate_csv N2 F G 2 0.976 0.2273 0.7487 1
athanasius_epistles_data_multistate_csv N4 M223 M2423 2 0.944 0.3298 0.6142 1
athanasius_epistles_data_multistate_csv N5 D F G 3 0.777333 0.203111 0.574222 0.833333
athanasius_epistles_data_multistate_csv N3 M1739 U1 2 0.784 0.407 0.377 1
athanasius_epistles_data_multistate_csv N6 A B C M1739 P46 U1 6 0.612 0.287667 0.324333 0.833333
athanasius_epistles_data_multistate_csv N1 B C 2 0.746 0.431 0.315 1
athanasius_epistles_data_multistate_csv N7 A Ath B C M1739 M223 M2423 P46 U1 9 0.494278 0.203111 0.291167 0.833333