https://peterponyu.github.io/scCCVGBen/metrics/Recent content in Metrics on scCCVGBen — single-cell scCCVGBen BenchmarkHugo -- gohugo.ioen-ushttps://peterponyu.github.io/scCCVGBen/metrics/asw/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/asw/ASW # Field Value Group Clustering quality Source sklearn.metrics.silhouette_score Direction Higher is better Description # Average Silhouette Width Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/dav/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/dav/DAV # Field Value Group Clustering quality Source sklearn.metrics.davies_bouldin_score Direction Lower is better Description # Davies–Bouldin Index Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/cal/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/cal/CAL # Field Value Group Clustering quality Source sklearn.metrics.calinski_harabasz_score Direction Higher is better Description # Calinski–Harabasz Index Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/distance-correlation-umap/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/distance-correlation-umap/distance_correlation_umap # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE.evaluate_dimensionality_reduction Direction — Description # Spearman distance corr (latent vs UMAP) Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/q-local-umap/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/q-local-umap/Q_local_umap # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE Lee & Verleysen 2009 Direction — Description # Coranking Q_NX local average over the selected neighbourhood range Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/q-global-umap/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/q-global-umap/Q_global_umap # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE Direction — Description # Coranking Q_NX global average Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/k-max-umap/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/k-max-umap/K_max_umap # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE Direction — Description # Argmax of Q_NX(k) — optimal neighbourhood scale Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/overall-quality-umap/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/overall-quality-umap/overall_quality_umap # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE Direction — Description # Weighted composite of Q_local + Q_global Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/distance-correlation-tsne/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/distance-correlation-tsne/distance_correlation_tsne # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE Direction — Description # Spearman dist corr (latent vs tSNE) Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/q-local-tsne/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/q-local-tsne/Q_local_tsne # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE Direction — Description # Coranking Q_NX local (tSNE) Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/q-global-tsne/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/q-global-tsne/Q_global_tsne # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE Direction — Description # Coranking Q_NX global (tSNE) Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/k-max-tsne/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/k-max-tsne/K_max_tsne # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE Direction — Description # Argmax Q_NX(k) in tSNE Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/overall-quality-tsne/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/overall-quality-tsne/overall_quality_tsne # Field Value Group Dimensionality-reduction evaluation (UMAP + tSNE) Source DRE Direction — Description # Weighted composite (tSNE) Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/manifold-dimensionality-intrin/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/manifold-dimensionality-intrin/manifold_dimensionality_intrin # Field Value Group Latent-space intrinsic geometry Source LSE.SingleCellLatentSpaceEvaluator Direction — Description # Intrinsic manifold dimensionality score Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/spectral-decay-rate-intrin/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/spectral-decay-rate-intrin/spectral_decay_rate_intrin # Field Value Group Latent-space intrinsic geometry Source LSE Direction — Description # PCA eigenvalue decay rate Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/participation-ratio-intrin/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/participation-ratio-intrin/participation_ratio_intrin # Field Value Group Latent-space intrinsic geometry Source LSE Direction — Description # Effective latent dim (1/Σ w_i² when Σ w_i=1) Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/anisotropy-score-intrin/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/anisotropy-score-intrin/anisotropy_score_intrin # Field Value Group Latent-space intrinsic geometry Source LSE Direction — Description # Covariance eigenvalue anisotropy Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/trajectory-directionality-intrin/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/trajectory-directionality-intrin/trajectory_directionality_intrin # Field Value Group Latent-space intrinsic geometry Source LSE Direction — Description # Primary developmental axis strength Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/noise-resilience-intrin/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/noise-resilience-intrin/noise_resilience_intrin # Field Value Group Latent-space intrinsic geometry Source LSE Direction — Description # Perturbation robustness score Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.https://peterponyu.github.io/scCCVGBen/metrics/overall-quality-intrin/Mon, 01 Jan 0001 00:00:00 +0000https://peterponyu.github.io/scCCVGBen/metrics/overall-quality-intrin/overall_quality_intrin # Field Value Group Latent-space intrinsic geometry Source LSE Direction — Description # Weighted composite (trajectory-aware) Every method’s latent embedding is scored against this metric across all 200 benchmark datasets; see the Metrics index for the full set.