Pith. sign in

REVIEW 3 major objections 5 minor 27 references

cGAP embeds subjects and category levels in a 3D space and colors each cell of a categorical data matrix so that similar response patterns receive similar colors.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

cGAP colors and reorders a categorical data matrix using a 3D HOMALS embedding, producing coordinated heatmap and proximity views for exploratory analysis.

T0 review reviewed 2026-08-02 challenge →

load-bearing objection A useful categorical extension of GAP with an honest but flawed formal derivation: Eq. (5) is wrong as written, and the key structure-discovery claims rest on visual inspection rather than measured evaluation. the 3 major comments →

arxiv 2607.15018 v2 pith:YUV6N7ZW submitted 2026-07-16 stat.ML cs.LGstat.COstat.ME

cGAP: Generalized Association Plots with HOMALS-Guided Heatmaps for Visualization of High-Dimensional Categorical Data

Chun-houh Chen , Shun-Chuan Chang , Chiun-How Kao , Yi-Ju Lee , Shang-Ying Shiu , Yin-Jing Tien , ShengLi Tzeng , Han-Ming Wu This is my paper
classification stat.ML cs.LGstat.COstat.ME
keywords matrix visualizationcategorical datahomogeneity analysisHOMALSseriationheatmapcolor encodingexploratory data analysis
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that high-dimensional categorical data can be visualized as a full-matrix heatmap whose cell colors come from a three-dimensional homogeneity-analysis (HOMALS) embedding, so that subjects with similar response profiles receive similar hues. The claim gains force from a barycentric traceability result: because HOMALS makes each subject the average of its selected categories, and the RGB relocation is affine, a subject's displayed color is literally the average of the colors of its categories. The framework adds subject- and variable-proximity panels and a seriation step, and it proves that projecting from the full HOMALS space to three dimensions only shortens distances, with the loss exactly the sum of squared discarded coordinates, and that the optional contrast transform preserves rays from the cube center. Applications to ordinal, nominal, and binary datasets show clusters, outliers, and inverse present/absent modules while keeping every original entry readable. If the central claim holds, analysts gain a matrix-centered categorical visualization that scales to large genomic data without sacrificing traceability to raw observations.

Core claim

cGAP uses homogeneity analysis (HOMALS) to embed subjects and the category levels of all variables into a common three-dimensional Euclidean space, then linearly relocates those coordinates into the unit RGB cube so each category receives a color and each subject receives, as its display color, the average of the colors of the categories in its response profile. The discovery is that this barycentric identity survives the color map exactly, so the colored matrix is not arbitrary paletting but a geometric representation of the embedding that remains traceable: every cell's color can be read as part of a subject profile that averages its categories. Two further results support the display: the

What carries the argument

The central machinery is the three-dimensional HOMALS embedding together with the affine relocation T(z)=z/(2m)+0.5 into the RGB cube. HOMALS—homogeneity analysis, an alternating least-squares method for jointly embedding subjects and category levels—is built on update equations that make subject coordinates the average of their selected category coordinates, and it is this barycentric identity that gives cGAP its traceability: after the affine color map, a subject's displayed color is the average of its category colors. The two supporting identities are the projection-distortion decomposition, which separates the displayed squared distance between two points into the first-p contribution pl

Load-bearing premise

The interpretability of the display rests on the assumption that Euclidean distance in the three-dimensional HOMALS embedding, after affine rescaling into the RGB cube, is a faithful proxy for the color similarity a viewer perceives; if that link fails, the claim that similar patterns receive similar colors becomes a display artifact.

What would settle it

Take a dataset with a known HOMALS embedding, compute the displayed RGB colors for all subjects, and ask human observers (or a perceptual color-difference formula such as CIEDE2000) to rank the dissimilarity of selected pairs. If the perceived or formula-based color dissimilarity does not monotonically track the HOMALS distances—especially for pairs that are close in the embedding but map to grayish versus saturated regions of the cube—the central claim is falsified. A simpler, purely computational symptom: find two subject pairs with equal HOMALS separation located in different parts of the R

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • cGAP keeps the raw data matrix as the primary display, so users can inspect any individual cell while still benefiting from embedding-derived color similarity and seriation.
  • Because subject colors are affine averages of their category colors, a subject with an odd response profile receives a mixed, atypical color, making observations such as the alligator or armadillo immediately visible.
  • The induced variable dissimilarity satisfies symmetry, identity, and the triangle inequality, so variables can be ordered and clustered directly on embedding geometry—recovering, for example, the pairing of upper and lower tooth variables in dentition data.
  • On large binary genomic data, the method exposes complementary present/absent modules and near-universal cores without collapsing to simple presence counts, because both states are embedded and colored jointly.
  • The projection-distortion decomposition gives a quantitative check on how much geometry is lost in three dimensions, telling a user when to trust color proximity and when to fall back on the raw matrix.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The barycentric color identity would survive any invertible (or monotone) map from the three embedding coordinates to color, so switching to a perceptually uniform space such as CIELAB could make 'similar patterns, similar colors' true in a visual sense and is a natural extension the paper leaves open.
  • Because the paper reports retained variation only for the dentition example (about 67%), a simple discipline for users would be to report the three-dimensional retention rate for every cGAP figure and treat color-based conclusions as provisional when it is low.
  • The authors note that very large matrices may need interactive or supplementary views; a concrete next step is a linked-brush interface in which color encoding, proximity panels, and seriation respond to selections, turning the traceability theorem into a navigation tool.
  • The load-bearing color-proximity link is empirically testable: present observers with pairs of colored cells from a known embedding and compare their judged similarity with HOMALS distances; disagreements, especially in gray versus saturated regions, would locate where RGB fails as a perceptual proxy.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes cGAP, a visualization framework for high-dimensional categorical data. It uses homogeneity analysis (HOMALS) to embed subjects and category levels in a three-dimensional Euclidean space, maps those coordinates to RGB color channels, and constructs three coordinated views: a raw-data heatmap with embedding-derived colors, a subject proximity matrix, and a variable proximity matrix. Row and column seriation (R2E, R1T, HCT-R2E) is then applied to reorder the displays to reveal clusters, outliers, and association structure. The paper also states several formal properties of the embedding, including barycentric traceability, a projection-distortion decomposition, and a ray-preserving contrast transform. The method is illustrated on four datasets: student-animal grouping, mammalian dentition, mushroom attributes, and COG genomic profiles.

Significance. If the framework is effective, it would be a useful addition to categorical data visualization, preserving the original data matrix while adding embedding-derived geometric structure. The paper provides a concrete implementation (publicly available software) and introduces a specific pipeline that goes beyond traditional low-dimensional scatterplots. The mathematical derivations are mostly elementary, but they serve to justify the interpretability of the displays. The main weakness is that the central claim—that the colors and proximity views reveal structure—is only supported by subjective visual inspection; no quantitative evaluation or comparison against existing tools is provided. As a methods paper, the practical contribution is plausible but not yet demonstrated to the usual standard for a statistics/machine-learning journal.

major comments (3)
  1. [Section 2.2, Eq. (5)] Equation (5) is algebraically incorrect. For optimal category coordinates Y_j = (G_j^t G_j)^{-1} G_j^t X, the per-variable residual is tr((X - G_j Y_j)^t (X - G_j Y_j)) = Np - N \sum_s \eta^2_{js}, because \eta^2_{js} is normalized by N. Averaging over J variables gives \sigma = Np - (N/J) \sum_j \sum_s \eta^2_{js} = Np - N \sum_s \gamma_s, not Np - \sum_j \sum_s \eta^2_{js} as written. This error does not change the ratio \gamma^*_p because the N/J factor cancels, but the stated loss identity is false and cannot serve as the formal basis for the claim that \gamma^*_p measures the proportion of retained variation. Please correct the equation and the surrounding derivation.
  2. [Section 4, applications] The central claim that cGAP 'reveals' clusters, outliers, and association structure is supported only by visual inspection of the four examples. There is no quantitative measure of cluster recovery (e.g., agreement of the seriated blocks with known taxonomic or class labels), no comparison against alternative categorical displays (e.g., MCA scatterplot, textile plot, or a heatmap with a trivial color scheme), and no user study of interpretability. Because the paper positions itself as improving interpretability over existing methods, this lack of evidence is load-bearing. Please add at least one quantitative evaluation, such as measuring the accuracy of the identified clusters against known groupings, or systematically comparing color-distance agreement with HOMALS-distance against a random-color baseline.
  3. [Section 3.1, color encoding] The claim that 'similar patterns receive similar colors' assumes that Euclidean distance in the RGB cube is a faithful proxy for perceived color similarity. The paper acknowledges that RGB is not perceptually uniform, yet the interpretation of the displays throughout Section 4 relies on it. This is a potential gap in the logic: the affine mapping preserves geometry, but the perceptual relevance of that geometry is not established. I recommend either using a perceptually uniform color space, providing evidence that the RGB mapping preserves the perceptual ordering, or explicitly limiting the claim to 'geometric proximity' rather than 'perceived color similarity'.
minor comments (5)
  1. [Corollary 1] The relocation map is written as T(z) = z/(2m) + 0.513; the '0.513' appears to be a typo for 0.5 * 1_3. Please clarify.
  2. [Section 3.5, contrast transform] The inline formula for z^{(q)}(k,r) is difficult to read and appears garbled. Proposition 3 implies z^{(q)}(k,:) - 0.5 1_3 = w_k^{1/q - 1}(z(k,:) - 0.5 1_3). Please rewrite the formula cleanly and check the q=1 case.
  3. [Section 3.3, variable proximity] The proof of the identity-of-indiscernibles condition is not fully correct: two distinct category labels can have identical HOMALS coordinates if their indicator columns coincide, so d(Z_k,Z_l)=0 does not necessarily imply Z_k = Z_l as functions. Since the paper already calls this a 'weighted embedding-based dissimilarity', consider softening the metric claim.
  4. [Section 4.1] The 67% retained variation for the dentition example should state explicitly that it is \gamma^*_3, and the same quantity should be reported for the other datasets if available.
  5. [General] The text contains several minor typographical and formatting issues, such as the inconsistent use of 'HOMALS' vs. 'Homogeneity Analysis' and the garbled equations in Section 3.5. A careful proofreading pass is recommended.

Circularity Check

0 steps flagged

No significant circularity: the derivation chain is a transparent construction from the HOMALS fit, not a prediction or a self-citation-forced result.

full rationale

cGAP is a visualization framework, not a predictive or inferential method. The colors (Eqs. 12–13), subject proximity (Sec. 3.2), and variable proximity (Eq. 14) are explicitly constructed from the fitted HOMALS coordinates; the paper does not present these views as independent validations of the embedding, nor as predictions of external outcomes. The 'revealed' structure is therefore by construction, but this is openly stated and is the intended role of the embedding-guided display. External anchors such as dentition taxonomy, mushroom edibility labels, and COG taxonomic categories are used only for post-hoc interpretation, not as fitted inputs. The formal propositions (barycentric traceability, projection-distortion, contrast transform) are algebraic identities within the HOMALS equations; they are tautological but are not load-bearing claims that the method works. Self-citations to Chen (2002), Tien et al. (2008), and Wu et al. (2010) supply design principles and seriation algorithms, not an unverified uniqueness theorem or a forbidden alternative. The algebraic slip in Eq. (5), which appears to drop an N/J factor, is a correctness issue in the derivation of the retained-variation measure, not a circularity: gamma* is an internal quality statistic and the reported proportion is scale-invariant. No fitted parameter is renamed as a prediction, and no load-bearing argument reduces to a self-citation chain. Thus no significant circularity is present.

Axiom & Free-Parameter Ledger

3 free parameters · 5 axioms · 0 invented entities

No new physical or statistical entities are postulated. The main un-imported choices are p=3, the contrast parameter q, and the ordinal-constraint specification; the main domain assumptions are that HOMALS Euclidean proximity and RGB distance are faithful proxies for structure and perception.

free parameters (3)
  • embedding dimension p=3 = 3
    Chosen to match RGB channels; affects color fidelity. Retained variation gamma*_3 is reported only for the dentition example (67%), not for mushrooms or COG.
  • contrast parameter q = not stated for figures
    Defined in Section 3.5 as a power transform on relocated coordinates; the actual q values used in Figures 4–8 are not reported, so exact displays are not reproducible.
  • ordinal constraints for HOMALS = unspecified
    Section 4.1 says HOMALS is fit "with order constraints" but does not specify the constraints or implementation, affecting the dentition embedding and colors.
axioms (5)
  • standard math HOMALS alternating least squares converges to a meaningful optimal embedding (Gifi 1990; de Leeuw et al. 1967).
    cGAP's colors and proximities are taken from the HOMALS solution; correctness of the embedding algorithm is imported from cited literature.
  • domain assumption Euclidean distance in 3D HOMALS space reflects relevant categorical association structure.
    Sections 3.1–3.2; without this, color and subject-proximity panels would not reveal real clusters.
  • domain assumption RGB Euclidean distance is an acceptable stand-in for perceived color similarity.
    Section 3.1 acknowledges RGB is not perceptually optimal, yet the interpretation "similar colors = similar patterns" relies on color similarity being meaningful.
  • domain assumption Seriation (R2E/R1T/HCT-R2E) reorders rows and columns without creating spurious block structure.
    Used to produce sorted heatmaps in Sections 3.4 and 4; no simulation or null-model check is provided.
  • domain assumption External labels (taxonomy, edibility, COG categories) are appropriate ground truth for interpreting the displays.
    Applications read clusters against known labels; if labels are wrong or incomplete, the claimed validations weaken.

reviewed 2026-08-02 · how reviews work

0 comments
Cite this review

Pith. "Pith review of cGAP: Generalized Association Plots with HOMALS-Guided Heatmaps for Visualization of High-Dimensional Categorical Data." pith.science (2026). https://pith.science/paper/YUV6N7ZW

@misc{pith2026260715018,
  author       = {Pith},
  title        = {Pith review of: cGAP: Generalized Association Plots with HOMALS-Guided Heatmaps for Visualization of High-Dimensional Categorical Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YUV6N7ZW}},
  note         = {Machine review of arXiv:2607.15018}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

High-dimensional categorical data arise in genetics, biomedicine, and the social sciences, yet visualization tools for such data remain far less developed than those for continuous variables. Existing methods either scale poorly, rely heavily on low-dimensional displays detached from the original data matrix, or prioritize predictive accuracy over interpretability. To address this gap, we introduce categorical Generalized Association Plots (cGAP), a visualization framework for nominal, ordinal, and binary data that preserves the original data matrix while augmenting it with interpretable geometric structure. cGAP uses Homogeneity Analysis (HOMALS) to embed subjects and category levels in a three-dimensional Euclidean space and maps the embedding to red-green-blue coordinates so that similar patterns receive similar colors. The framework integrates three coordinated views: a HOMALS-guided heatmap of the raw data matrix, a subject proximity matrix, and a variable proximity matrix. Seriation algorithms are then used to reorder rows and columns to reveal coherent clusters, outliers, and local-to-global structure. We also derive barycentric traceability, projection-distortion, and contrast-preservation properties that clarify how embedding geometry is transferred to the display. We demonstrate the versatility of cGAP through applications to student-animal classification data, mammalian dentition profiles, mushroom records from the UCI Machine Learning Repository, and the Clusters of Orthologous Genes database. These examples show that cGAP supports transparent exploratory analysis by maintaining traceability between derived visual structure and the original categorical observations. cGAP provides a full-matrix, heatmap-based visualization environment for investigating complex categorical datasets across scientific domains.

Figures

Figures reproduced from arXiv: 2607.15018 by Chiun-How Kao, Chun-houh Chen, Han-Ming Wu, Shang-Ying Shiu, ShengLi Tzeng, Shun-Chuan Chang, Yi-Ju Lee, Yin-Jing Tien.

Figure 1
Figure 1. Figure 1: Graphical tools for continuous data and their categorical counterparts across [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a) The three-dimensional HOMALS embedding, where H1, H2, and H3 denote [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The unsorted cGAP display for the animal-grouping data in Table 1 (35 animal [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: cGAP with HCT-R2E seriation for the animal-grouping data in Table 1 (35 [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: cGAP analysis of the mammalian dentition data (66 mammals [PITH_FULL_IMAGE:figures/full_fig_p016_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: cGAP analysis of the mushroom data (8,124 mushrooms [PITH_FULL_IMAGE:figures/full_fig_p017_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: One sequence of interpretable rules, derived from the cGAP views, for dis [PITH_FULL_IMAGE:figures/full_fig_p018_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: cGAP analysis of the COG dataset (2,296 genomes [PITH_FULL_IMAGE:figures/full_fig_p019_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Sediment displays for the COG dataset (2,296 genomes [PITH_FULL_IMAGE:figures/full_fig_p020_9.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

27 extracted references · 2 canonical work pages

  1. [1]

    , year 1973

    author Benzecri, J.P. , year 1973 . title L'analyse des donnees. II. L'analyse des correspondances . publisher Dunod , address Paris

    1973

  2. [2]

    , year 2002

    author Chen, C.H. , year 2002 . title Generalized association plots: information visualization via iteratively generated correlation matrices . journal Statistica Sinica volume 12 , pages 7--29

    2002

  3. [3]

    , year 2004

    author Chen, C.H. , year 2004 . title Matrix visualization and information mining , in: booktitle COMPSTAT 2004 -- Proceedings in Computational Statistics , pp. pages 85--100

    2004

  4. [4]

    , author Asuncion, A

    author Frank, A. , author Asuncion, A. , year 2010 . title Uci machine learning repository . http://archive.ics.uci.edu/ml. note irvine, CA: University of California, School of Information and Computer Science

    2010

  5. [5]

    , year 1994

    author Friendly, M. , year 1994 . title Mosaic displays for multi-way contingency tables . journal Journal of the American Statistical Association volume 89 , pages 190--200

    1994

  6. [6]

    , year 1999

    author Friendly, M. , year 1999 . title Extending mosaic displays: Marginal, conditional, and partial views of categorical data . journal Journal of Computational and Graphical Statistics volume 8 , pages 373--395

    1999

  7. [7]

    , author Makarova, K.S

    author Galperin, M.Y. , author Makarova, K.S. , author Wolf, Y.I. , author Koonin, E.V. , year 2015 . title Expanded microbial genome coverage and improved protein family annotation in the COG database . journal Nucleic Acids Research volume 43 , pages D261--D269

    2015

  8. [8]

    , author Wolf, Y.I

    author Galperin, M.Y. , author Wolf, Y.I. , author Makarova, K.S. , author Vera Alvarez, R. , author Landsman, D. , author Koonin, E.V. , year 2021 . title COG database update: focus on microbial diversity, model organisms and widespread pathogens . journal Nucleic Acids Research volume 49 , pages D274--D281

    2021

  9. [9]

    , et al., year 2024

    author Galperin, M.Y. , et al., year 2024 . title COG database update 2024 . journal Nucleic Acids Research volume 53 , pages D356--D363 . https://doi.org/10.1093/nar/gkae983, :10.1093/nar/gkae983

    work page doi:10.1093/nar/gkae983 2024

  10. [10]

    , year 1990

    author Gifi, A. , year 1990 . title Nonlinear Multivariate Analysis . publisher John Wiley & Sons

    1990

  11. [11]

    , author Mushegian, A.R

    author Glazko, G.V. , author Mushegian, A.R. , year 2004 . title Detection of evolutionarily stable fragments of cellular pathways by hierarchical clustering of phyletic patterns . journal Genome Biology volume 5 , pages R32 . https://doi.org/10.1186/2004-5-5-r32, :10.1186/2004-5-5-r32

    work page doi:10.1186/2004-5-5-r32 2004

  12. [12]

    , year 1984

    author Greenacre, M.J. , year 1984 . title Theory and Applications of Correspondence Analysis . publisher Academic Press , address London

    1984

  13. [13]

    , year 1975

    author Hartigan, J. , year 1975 . title Clustering Algorithms . publisher Wiley , address New York, NY

    1975

  14. [14]

    , author Kleiner, B

    author Hartigan, J.A. , author Kleiner, B. , year 1981 . title Mosaics for contingency tables , in: editor Eddy, W.F. (Ed.), booktitle Computer Science and Statistics: Proceedings of the 13th Symposium on the Interface . publisher Springer-Verlag , address New York

    1981

  15. [15]

    , author Shibata, R

    author Kumasaka, N. , author Shibata, R. , year 2008 . title High-dimensional data visualisation: The textile plot . journal Computational statistics & data analysis volume 52 , pages 3616--3644

    2008

  16. [16]

    , author Young, F.W

    author de Leeuw, J. , author Young, F.W. , author Takane, Y. , year 1967 . title Additive structure in qualitative data: An alternating least squares method with optimal scaling features . journal Psychometrika volume 41 , pages 471--503

    1967

  17. [17]

    , author de Leeuw, J

    author Michailidis, G. , author de Leeuw, J. , year 1998 . title The Gifi system of descriptive multivariate analysis . journal Statistical Science volume 13 , pages 307--336

    1998

  18. [18]

    , author West, R.W

    author Minnotte, M. , author West, R.W. , year 1998 . title The Data Image : A tool for exploring high dimensional data sets , in: booktitle Proceedings of the Section on Statistical Graphics, American Statistical Association , address Alexandria, Virginia

    1998

  19. [19]

    , year 1984

    author Nishisato, S. , year 1984 . title Dual scaling of reciprocal medians , in: booktitle Proceedings of the 32nd Scientific Conference of the Italian Statistical Society , publisher Societa Italiana di Statistica , address Sorrento, Italy . pp. pages 141--147

    1984

  20. [20]

    , year 2007

    author Nishisato, S. , year 2007 . title Multidimensional Nonlinear Descriptive Analysis . publisher Chapman & Hall

    2007

  21. [21]

    , author Galperin, M.Y

    author Tatusov, R.L. , author Galperin, M.Y. , author Natale, D.A. , author Koonin, E.V. , year 2000 . title The COG database: a tool for genome-scale analysis of protein functions and evolution . journal Nucleic Acids Research volume 28 , pages 33--36

    2000

  22. [22]

    , author Koonin, E.V

    author Tatusov, R.L. , author Koonin, E.V. , author Lipman, D.J. , year 1997 . title A genomic perspective on protein families . journal Science volume 278 , pages 631--637

    1997

  23. [23]

    , author Natale, D.A

    author Tatusov, R.L. , author Natale, D.A. , author Garkavtsev, I.V. , author Tatusova, T.A. , author Shankavaram, U.T. , author Rao, B.S. , author Kiryutin, B. , author Galperin, M.Y. , author Fedorova, N.D. , author Koonin, E.V. , year 2001 . title The COG database: new developments in phylogenetic classification of proteins from complete genomes . jour...

    2001

  24. [24]

    , et al., year 2008

    author Tien, Y.J. , et al., year 2008 . title Methods for simultaneously identifying coherent local clusters with smooth global patterns in gene expression profiles . journal BMC Bioinformatics volume 9 , pages 1--16 . note Article 155

    2008

  25. [25]

    , author Wu, H.M

    author Tzeng, S.L. , author Wu, H.M. , author Chen, C.H. , year 2009 . title Selection of proximity measures for matrix visualization of binary data , in: booktitle Proceedings of the 2009 2nd International Conference on BioMedical Engineering and Informatics ( BMEI 2009) , pp. pages 1932--1940

    2009

  26. [26]

    , year 1990

    author Wegman, E.J. , year 1990 . title Hyperdimensional data analysis using parallel coordinates . journal Journal of the American Statistical Association volume 85 , pages 664--675

    1990

  27. [27]

    , author Tien, Y.J

    author Wu, H.M. , author Tien, Y.J. , author Chen, C.H. , year 2010 . title GAP : A graphical environment for matrix visualization and cluster analysis . journal Computational Statistics and Data Analysis volume 54 , pages 767--778

    2010

This paper was first reviewed by deepseek-v4-flash on August 2, 2026.