Pith. sign in

REVIEW 4 major objections 4 minor 34 references

Improving statistical learning methods via features selection without replacement sampling and random projection

T0 review · 4 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read A pipeline that combines Kendall gene filtering, random feature selection without replacement, and LDA projection reaches 96 percent test accuracy on a brain cancer microarray benchmark.

desk verdict The headline 96% accuracy is contradicted by the paper's own tables, and the feature-selection-before-CV leakage invalidates even the lower reported numbers. read the letter →

arxiv 2506.00053 v1 pith:B4I2QTCE submitted 2025-05-28 q-bio.QM cs.AIcs.LGstat.APstat.ML

classification q-bio.QMcs.AIcs.LGstat.APstat.ML
keywords braincancergeneexpressionmicroarrayfeatureselectionwithoutreplacementKendalltestrandomprojectionLDAensembleclassification
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that a three-stage pipeline—Kendall statistical testing to prune genes, random feature selection without replacement to build diverse views, and projection before classification—raises brain cancer subtype classification accuracy on the GSE50161 dataset to 96 percent. That would beat the previous benchmark on the same data by 9.09 percent. The authors test five classifiers under four projections and find Naïve Bayes with LDA projection is the winning configuration. Their case is that statistical preselection plus diversity from sampling without replacement solves the small-sample, high-dimensional overfitting problem.

What carries the argument

Feature Selection Without Replacement (FSWOR) is the central mechanism: features are drawn randomly without replacement to form distinct subsets, so each member of the ensemble trains on a different view of the data, which mitigates overfitting and adds diversity. Kendall's statistical test first reduces the gene space from 54,675 to 20,890 genes, and each feature subset is then projected using PCA, LDA, Gaussian random projection, or sparse random projection. LDA, which maximizes class separability, is the projection that carries the winning result. The experiment compares five classifiers (DT, SVM, LR, KNN, NB) under each projection using 3-fold cross-validation, with the ensemble combining the individual learners.

What would settle it

Recompute the pipeline with the Kendall test and FSWOR sampling performed only inside each training fold, and check whether the held-out test accuracy stays at 96 percent; a material drop would show the reported gain over the 88 percent baseline comes from selection leakage rather than the method itself.

Watch

Extended reading notes

Core claim

The central discovery is that feature selection without replacement, combined with Linear Discriminant Analysis projection, pushes a Naïve Bayes classifier to 0.938 accuracy (96 percent in the paper's reporting) on the brain cancer microarray dataset GSE50161, above the 0.88 previously reported for this benchmark. Under the same preprocessing, the same classifier reaches only 0.892 with PCA, 0.815 with Gaussian random projection, and 0.835 with sparse random projection, while SVM becomes the best model under PCA, GRP, and SRP. The paper attributes the gain to the ensemble diversity created by sampling features without replacement and to LDA's ability to make the five tissue classes linearly separable in the reduced space. Decision trees perform worst under random projections, with accuracy as low as 0.619 for SRP.

Load-bearing premise

The Kendall test and the feature-sampling step are applied to the full dataset before the cross-validation folds are split, so the reported accuracy assumes that using all samples for feature selection does not leak information into the held-out folds.

Editorial extensions

If this is right

  • On the GSE50161 dataset, the proposed FSWOR+LDA+Naïve Bayes setup reaches 96 percent cross-validation accuracy, a 9.09 percent relative gain over the prior 88 percent benchmark.
  • LDA is the decisive projection: Naïve Bayes with LDA scores 0.938, versus 0.892 with PCA, 0.815 with GRP, and 0.835 with SRP.
  • SVM is the most consistent classifier across projections, with accuracy 0.873-0.915, while decision trees degrade sharply under random projections.
  • The Kendall test cuts the feature space from 54,675 to 20,890 genes, reducing dimensionality before any model is trained.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An untested but direct follow-up is to wrap the Kendall test and FSWOR inside each cross-validation fold; that would tell whether the 96 percent figure survives data leakage from the feature-selection step.
  • The dominance of LDA over random projections on this dataset suggests the five tissue classes are close to linearly separable after gene pruning, so the FSWOR ensemble's diversity may matter less than the projection choice.
  • On datasets where the signal is spread across many weak features rather than concentrated in a few strong genes, random projections might overtake LDA; comparing the two on such data would test the generality of the claim.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes a supervised learning pipeline for brain cancer microarray classification (GSE50161). The method first applies a Kendall statistical test to the full dataset to reduce the feature space from 54,675 to 20,890 genes, then constructs ensemble classifiers through feature selection without replacement (FSWOR) and four projection techniques (PCA, LDA, GRP, SRP), evaluated with 3-fold cross-validation. The central claims are that the LDA-projection NB ensemble achieves a test accuracy of 96%, outperforming prior work by 9.09%.

Significance. If the claims were validated, the paper would offer a useful empirical benchmark for high-dimensional gene-expression classification. The work uses a publicly available dataset and presents a clearly structured pipeline. However, the reported headline accuracy is contradicted by the paper's own tables, the validation protocol leaks test information into feature selection, and the confusion-matrix descriptions contain impossible percentages. As a result, the central empirical contribution is currently unsupported.

major comments (4)
  1. [Abstract, Table 1, Table 3, Table 6] The abstract and Table 1 claim a test score of 96% and a 9.09% improvement over prior methods, but Tables 3 and 6 report the best accuracy as 0.938 for Naïve Bayes with LDA. The stated 9.09% improvement appears to be computed as 0.96/0.88, not as 0.938/0.88 ≈ 6.6%. The headline number is therefore not derivable from the reported experimental results.
  2. [Sections 3.1 and 4.1] The Kendall statistical test is applied to the complete dataset before any train/test split, and the 20,890-gene subset is selected using all 130 samples. Because the evaluation uses 3-fold cross-validation, each held-out fold has already contributed to gene ranking and normalization. This is a standard feature-selection leakage path that inflates the cross-validated accuracy, so the comparison with the baseline is not a fair estimate of generalization performance.
  3. [Figure 12 and Section 4.2] The confusion-matrix descriptions in Figure 12 contain percentages that exceed 100% (e.g., 590% correct for class 1, 220% misclassified, 110%, 900%, 870%) and are internally inconsistent across classes. If these values are literal, they indicate severe numerical errors; if they are unnormalized counts mislabeled as percentages, the reporting is misleading. Either way, the detailed per-class performance claims cannot be accepted.
  4. [Table 6 and Section 4.2] The paper evaluates five classifiers under four projections (20 combinations) and then selects, post hoc, the best model per projection for the summary in Table 6. With no separate validation set and no adjustment for multiple comparisons, the reported best accuracy (0.938) is optimistically biased and does not represent a pre-specified model's expected test performance.
minor comments (4)
  1. [Throughout] The text contains numerous typographical errors and formatting inconsistencies, including “project ion”, “Naïve Bays”, “violets Gaussian Markov assumption” (should be “violates”), and inconsistent spacing in numbers such as “20, 890” and “54, 675”.
  2. [Section 3.2] The FSWOR technique is described only qualitatively; no algorithm, pseudocode, or precise definition of the feature-subset sizes or sampling strategy is provided, which limits reproducibility.
  3. [Figures 8–11] The captions refer to accuracy versus “number of projections,” but the manuscript does not state the ranges of projection dimensions tested, the stopping criterion, or how the optimal dimension was chosen.
  4. [Table 1 and References] The comparison baseline is attributed to “Bruno et al.” in the introduction, but the reference list contains Feltes et al. [1]; the source of the 0.88 CV score is not clearly identified.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is an empirical pipeline study with no derivation chain whose output reduces to its inputs.

full rationale

The paper is an empirical pipeline study rather than a derivation: no result is derived from a theorem, and no fitted parameter is renamed as a prediction. The FSWOR step is described as random feature selection within ensembles, the Kendall test is applied as a preprocessing filter, and the projection/classifier combinations are standard building blocks with no definitional interdependence. The central 96% claim is not present in the reported results Tables 2-6 (the best reported accuracy is 0.938 for NB+LDA in Table 3), and the '9.09%' improvement corresponds to the 0.96 versus 0.88 values in Table 1 rather than to any experiment in the paper. This is an internal inconsistency and a statistical validity concern, not circular reasoning. Similarly, applying Kendall feature selection before the 3-fold split creates potential information leakage that may inflate the accuracy estimate, but leakage is a methodological flaw, not a self-referential derivation. There are no load-bearing self-citations, no imported uniqueness theorems, and no ansatz smuggled in via citation. The comparison to Feltes et al. is an external benchmark, not an input. Accordingly, no circular step can be quoted with a specific reduction, and the circularity score is 0.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The central claim depends on several unstated choices: the Kendall test threshold, the FSWOR ensemble parameters, projection dimensions, and random seeds. The analysis also relies on the assumption that filtering genes before cross-validation is unbiased, which is likely violated. No new entities are introduced.

free parameters (5)
  • Kendall test significance threshold = not reported
    Determines the number of selected genes (20,890 from 54,675); the threshold is not stated, so the feature set is a free choice.
  • Number and size of FSWOR feature subsets
    The ensemble behavior depends on how many random feature subsets are drawn and how large they are; these parameters are not specified.
  • Projection dimensions for PCA/LDA/GRP/SRP
    Figures 8-11 vary the number of projections, but the reported accuracy values do not state which dimensions were used for the final results.
  • Random seed
    FSWOR and random projections are stochastic; no seed is provided, so the reported results are not reproducible.
  • Number of cross-validation folds = 3
    The choice of 3-fold CV is fixed by design and affects the variance and bias of the accuracy estimate, especially with only 130 samples.
assumptions (3)
  • domain assumption Univariate Kendall association with class labels identifies genes useful for classification
    The paper assumes genes with low Kendall p-values are the most discriminative features. This is a standard filter assumption but is not guaranteed to hold for classification.
  • ad hoc to paper Feature selection can be performed on the full dataset before cross-validation without bias
    The pipeline applies the Kendall test before splitting into folds (Sections 3.1 and 4.1), which leaks test information into the feature selection step and inflates accuracy.
  • domain assumption Accuracy is a meaningful metric for this imbalanced dataset
    The class distribution is imbalanced (13 normal vs 117 abnormal), yet the headline uses accuracy alone. Per-class metrics are needed to avoid misleading conclusions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Improving statistical learning methods via features selection without replacement sampling and random projection." pith.science (2026). https://pith.science/paper/B4I2QTCE

@misc{pith2026250600053,
  author       = {Pith},
  title        = {Pith review of: Improving statistical learning methods via features selection without replacement sampling and random projection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/B4I2QTCE}},
  note         = {Machine review of arXiv:2506.00053}
}
read the original abstract

Cancer is fundamentally a genetic disease characterized by genetic and epigenetic alterations that disrupt normal gene expression, leading to uncontrolled cell growth and metastasis. High-dimensional microarray datasets pose challenges for classification models due to the "small n, large p" problem, resulting in overfitting. This study makes three different key contributions: 1) we propose a machine learning-based approach integrating the Feature Selection Without Re-placement (FSWOR) technique and a projection method to improve classification accuracy. 2) We apply the Kendall statistical test to identify the most significant genes from the brain cancer mi-croarray dataset (GSE50161), reducing the feature space from 54,675 to 20,890 genes.3) we apply machine learning models using k-fold cross validation techniques in which our model incorpo-rates ensemble classifiers with LDA projection and Na\"ive Bayes, achieving a test score of 96%, outperforming existing methods by 9.09%. The results demonstrate the effectiveness of our ap-proach in high-dimensional gene expression analysis, improving classification accuracy while mitigating overfitting. This study contributes to cancer biomarker discovery, offering a robust computational method for analyzing microarray data.

Figures

Figures reproduced from arXiv: 2506.00053 by the authors.

Figure 4
Figure 4. represent the results of a Kendall Statistical Test, which is a non-parametric test used to assess the association between two measured quantities. The y-axis likely rep￾resents the p-values obtained from the test, with values ranging from 5.00E-22 (a very small number indicating a highly significant result) down to 0.00E+00. The x-axis lists dif￾ferent identifiers, possibly representing genes or probes, such as "23… view at source ↗
Figure 5
Figure 5. Scatter plot after Kendall Statistical Test [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Histogram of Genes 4.2 Machine learning analysis 4.2.1 Classifier performance [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figures from the paper (1 more)
Figure 7
Figure 7. Figure 7: Comparisons of different machine learning models using different projection tech￾niques [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

34 extracted references · 33 canonical work pages

  1. [1]

    These alterations disrupt normal gene expression patterns [3], affecting genes that regulate cell growth, survival, and invasion while suppressing growth-inhibiting genes

    Introduction Cancer is fundamentally a genetic disease resulting from genetic or epigenetic alter- ations within somatic cells. These alterations disrupt normal gene expression patterns [3], affecting genes that regulate cell growth, survival, and invasion while suppressing growth-inhibiting genes. The primary mechanism involves the accumulation of muta- ...

  2. [2]

    [21] proposed a random projection enhancement (RPE) method to im- prove surrogate model performance

    Literature review Zhang et al. [21] proposed a random projection enhancement (RPE) method to im- prove surrogate model performance. They applied RPE to least squares support vector regression (LSSVR) and conducted numerical experiments. The results showed improved predictive accuracy, robustness, and optimization performance, even for high -dimen- sional ...

  3. [3]

    Methodology 3.1. Brain Cancer Dataset We sourced our dataset from Kaggle (https://www.kaggle.com/datasets/bruno- grisci/brain-cancer-gene-expression-cumida), which contains gene expression data for brain cancer. The dataset comprises a total of 130 samples, with 13 classes as normal and 117 classes as abnormal tissues. In the abnormal cases, there are fou...

  4. [4]

    230763_at,

    Results and Analysis 4.1 Statistical Analysis We have mention class distribution, scatterplot before Kendall statistical test and af- ter Kendall statistical test, obtained the most significant features i.e 20890. These features are used in Machine learning models with total samples 130. Figure 2 describe s different classes of Brain cancer type. Ependymo...

  5. [5]

    We applied the Kendall s tatistical test to identify 20, 890 significant genes from total genes 54,675 across 130 samples

    Conclusion In this study, we proposed a machine learning -based model utilizing the FSWOR technique and a projection method to address overfitting in high -dimensional brain can- cer microarray data (GSE50161). We applied the Kendall s tatistical test to identify 20, 890 significant genes from total genes 54,675 across 130 samples. Our approach integrated...

  6. [6]

    C., Chandelier, E

    Feltes, B. C., Chandelier, E. B., Grisci, B. I., & Dorn, M. (2019). Cumida: An extensively curated microarray database for bench- marking and testing of machine learning approaches in cancer research. Journal of Computational Biology, 26(4), 376-386

  7. [7]

    Brownlee, J. (2020). Data preparation for machine learning: data cleaning, feature selection, and data transforms in Python. Machine Learning Mastery

  8. [8]

    Nenclares, P., & Harrington, K. J. (2020). The biology of cancer. Medicine, 48(2), 67-72

Show all 34 references
  1. [9]

    Wang, Y., Liu, C., Fang, C., Peng, Q., Qin, W., Yan, X., & Zhang, K. (2025). Engineered Cancer Nanovaccines: A New Frontier in Cancer Therapy. Nano-Micro Letters, 17(1), 30

  2. [10]

    Y., & Liou, C

    Hsu, P. Y., & Liou, C. F. (2025). Impact of patient resourcefulness on cancer patients’ pain management and medical opioid use: A cross-sectional study. European Journal of Oncology Nursing, 74, 102771

  3. [11]

    Lu, X., Jin, J., Wu, Y., Lin, J., Zhang, X., Lu, S. ... & Luan, X. (2025). Self -assembled PROTACs enable protein degradation to reprogram the tumor microenvironment for synergistically enhanced colorectal cancer immunotherapy. Bioactive Materials, 43, 255-272

  4. [12]

    N., & Alagarsamy, M

    Murugan, K., Dhi vya, R., Sangeetha, C. N., & Alagarsamy, M. (2025). Design and Analysis of a Highly Sensitive Terahertz Biosensor for Early Cancer Detection Using Silver Surface Plasmon Resonance Metasurfaces and Elastic Reflection Starling Murmuration Equivariant Quantum Dec...

  5. [13]

    & Liu, J

    Cai, G., Huang, F., Gao, Y., Li, X., Chi, J., Xie, J., ... & Liu, J. (2024). Artificial intelligence -based models enabling accurate diag- nosis of ovarian cancer using laboratory tests in China: a multicentre, retrospective cohort study. The Lancet Digital health, 6(3), e176-e186

  6. [14]

    S., Aimetti, M., Sanz, M., Loos, B

    Baima, G., Minoli, M., Michaud, D. S., Aimetti, M., Sanz, M., Loos, B. G., & Romandini, M. (2024). Periodontitis and risk of cancer: Mechanistic evidence. Periodontology 2000, 96(1), 83-94

  7. [15]

    H., Hosseininia, H

    Abdolahi, M., Ghaedi Talkhounche, P., Derakhshan Nazari, M. H., Hosseininia, H. S., Khoshdel-Rad, N., & Ebrahimi Sadrabadi, A. (2024). Functional Enrichment Analysis of Tumor Microenvironment –Driven Molecular Alterations That Facilitate Epithe- lial-to-Mesenchymal Transition ...

  8. [16]

    H., Yahyazadeh, S., Irani, A., Banihashemi, S.,

    Liaghat, M., Ferdousmakan, S., Mortazavi, S. H., Yahyazadeh, S., Irani, A., Banihashemi, S., ... & Nabi -Afjadi, M. (2024). The impact of epithelial -mesenchymal transition (EMT) induced by metabolic processes and intracellular signaling pathways on chemo-resistance, metastasi...

  9. [17]

    Agraval, H., Kandhari, K., & Yadav, U. C. (2024). MMPs as potential molecular target in epithelial -to-mesenchymal transition driven COPD progression. Life Sciences, 122874

  10. [18]

    Schiliro, C., & Firestein, B. L. (2021). Mechanisms of metabolic reprogramming in cancer cells supporting enhanced growth and proliferation. Cells, 10(5), 1056

  11. [19]

    Holá, A. (2023). English for Medical Students. Charles University in Prague, Karolinum Press

  12. [20]

    Ishibashi, K., & Hirata, E. (2024). Multifaceted interactions between cancer cells and glial cells in brain metastasis. Cancer Sci- ence, 115(9), 2871-2878

  13. [21]

    Visentin, L., Scarpellino, G., Chinigò, G., Munaron, L., & Ruffinatti, F. A. (2022). BioTEA: containerized methods of analysis for microarray-based transcriptomics data. Biology, 11(9), 1346

  14. [22]

    Jovic, D., Liang, X., Zeng, H., Lin, L., Xu, F., & Luo, Y. (2022). Single‐cell RNA sequencing technologies and applications: A brief overview. Clinical and translational medicine, 12(3), e694

  15. [23]

    Zanella, L., Facco, P., Bezzo, F., & Cimetta, E. (2022). Feature selection and molecular classification of cancer phenotypes: a comparative study. International journal of molecular sciences, 23(16), 9087

  16. [24]

    A., Gaitsch, H., Wicks, E

    Caraway, C. A., Gaitsch, H., Wicks, E. E., Kalluri, A., Kunadi, N., & Tyler, B. M. (2022). Polymeric nanoparticles in brain cancer therapy: A review of current approaches. Polymers, 14(14), 2963

  17. [25]

    Abdel-Basset, M., Ding, W., & El -Shahat, D. (2021). A hybrid Harris Hawks optimization algorithm with simulated annealing for feature selection. Artificial Intelligence Review, 54(1), 593-637

  18. [26]

    Zhang, S., Pang, Y., Liu, F., Wang, M., Kan, Z., & Song, X. (2024). Random projection enhancement: A Novel method for im- proving performance of surrogate models. Swarm and Evolutionary Computation, 89, 101645

  19. [27]

    Hu, J., Hooi, B., & He, B. (2024). Efficient heterogeneous graph learning via random projection. IEEE Transactions on Knowledge and Data Engineering

  20. [28]

    Fabiani, G. (2024). Random projection neural networks of best approximation: Convergence theory and practical applications. arXiv preprint arXiv:2402.11397

  21. [29]

    Li, Y. (2024). Probability Tools for Sequential Random Projection. arXiv preprint arXiv:2402.14026

  22. [30]

    Asi, H., Feldman, V., Nelson, J., Nguyen, H., & Talwar, K. (2024). Fast optimal locally private mean estimation via random projections. Advances in Neural Information Processing Systems, 36

  23. [31]

    D., Gong, D., Parvaneh, A., Abbasnejad, E., & van den Hengel, A

    McDonnell, M. D., Gong, D., Parvaneh, A., Abbasnejad, E., & van den Hengel, A. (2024). Ranpac: Random projections and pre- trained models for continual learning. Advances in Neural Information Processing Systems, 36

  24. [32]

    Kumaran, K., Sajjan, M., Oh, S., & Kais, S. (2024). Random projection using random quantum circuits. Physical Review Research, 6(1), 013010

  25. [33]

    Social Support Detection from Social Media Texts

    Ahani, Zahra, et al. "Social Support Detection from Social Media Texts." arXiv preprint arXiv:2411.02580 (2024)

  26. [34]

    Analyzing hope speech from psycholinguistic and emotional perspectives

    Arif, Muhammad, et al. "Analyzing hope speech from psycholinguistic and emotional perspectives." Scientific re- ports 14.1 (2024): 23548. Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are s olely those of the individual au- thor(s...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.