pith. machine review for the scientific record. sign in

arxiv: 2605.00056 · v1 · submitted 2026-04-29 · 💻 cs.LG · cs.AI· physics.data-an· physics.geo-ph· stat.AP· stat.ML

Recognition: unknown

Smart Ensemble Learning Framework for Predicting Groundwater Heavy Metal Pollution

T. Ansah-Narh , G. Y. Afrifa , J. B. Tandoh , K. Asare , M. Addi , K. E. Yorke , D. M. A. Akpoley , K. Aidoo
show 1 more author
S. K. Fosuhene
Authors on Pith no claims yet

Pith reviewed 2026-05-09 19:49 UTC · model grok-4.3

classification 💻 cs.LG cs.AIphysics.data-anphysics.geo-phstat.APstat.ML
keywords groundwater pollutionheavy metal indexGaussian copulaensemble learningmachine learning predictionspatial mappingcontamination assessment
0
0 comments X

The pith

Gaussian copula transformation with stacked ensembles produces reliable high-accuracy maps of groundwater heavy metal pollution.

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

The paper sets out to demonstrate that standard machine learning models overfit when applied directly to the skewed and correlated Heavy Metal Pollution Index, but a Gaussian copula that normalizes those values first allows a stacked ensemble to deliver stable predictions. The authors test this on Densu Basin data by comparing raw, log, and copula versions across six learners and nested cross-validation. The copula version stands out by avoiding inflated fits, yielding better residuals, and generating maps that align with local chemistry. A sympathetic reader would care because accurate pollution forecasts matter for identifying real risks to drinking water supplies and guiding targeted remediation.

Core claim

The paper claims that transforming the Heavy Metal Pollution Index via Gaussian copula and feeding it into a stacked Lasso ensemble produces an R² of 0.96 and RMSE of 0.19, with other learners also showing strong performance, improved residual patterns, and spatially plausible output maps, while raw-scale models reach near-perfect R² values that mask over-optimism and log-scale models sit slightly lower.

What carries the argument

The Gaussian copula transformation that converts the skewed and inter-correlated Heavy Metal Pollution Index values into approximately normal form before they enter the stacked ensemble learner.

Load-bearing premise

Random nested cross-validation is enough to control for spatial patterns and correlations in the groundwater measurements.

What would settle it

Running the identical models with spatial block cross-validation and seeing the top R² fall substantially below 0.9 would undermine the reliability of the reported accuracy.

Figures

Figures reproduced from arXiv: 2605.00056 by D. M. A. Akpoley, G. Y. Afrifa, J. B. Tandoh, K. Aidoo, K. Asare, K. E. Yorke, M. Addi, S. K. Fosuhene, T. Ansah-Narh.

Figure 2
Figure 2. Figure 2 [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3 [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figure 3
Figure 3. Figure 3 [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figure 3
Figure 3. Figure 3 [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗
Figure 3
Figure 3. Figure 3 [PITH_FULL_IMAGE:figures/full_fig_p022_3.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p028_6.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p029_6.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p030_6.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p031_6.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p032_6.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p033_6.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p035_6.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p038_6.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p041_6.png] view at source ↗
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p042_6.png] view at source ↗
read the original abstract

Groundwater in the Densu Basin is increasingly threatened by heavy metal contamination, but conventional methods fail to capture the statistical complexity and spatial heterogeneity of pollution indicators. A key challenge is modelling the Heavy Metal Pollution Index (HPI), which is typically skewed and affected by correlated contaminants, leading to biased predictions without transformation. This study develops a predictive framework integrating response transformations with nested cross-validated ensemble machine learning. Three transformations (raw, log, and Gaussian copula) were applied to HPI and evaluated across six learners: support vector regression (SVM), $k$-nearest neighbours (k-NN), CART, Elastic Net, kernel ridge regression, and a stacked Lasso ensemble. Raw-scale models produced deceptively high fits (Elastic Net and stacked ensemble $R^2 \approx 1.0$), suggesting over-optimism. The log transformation stabilised variance (SVM: $R^2 = 0.93$, RMSE $= 0.18$; k-NN: $R^2 = 0.92$, RMSE $= 0.20$). The Gaussian copula gave the most reliable results: stacked ensemble $R^2 = 0.96$ (RMSE $= 0.19$), with other learners maintaining high accuracy. Copula-based models improved residuals and produced spatially plausible maps. DBSCAN clustering revealed Fe and Mn as primary HPI contributors, consistent with regional hydrogeochemistry. Limitations include reliance on random (not spatial) cross-validation and basin-specific scope. Future work should explore spatial validation and other geological settings. Overall, distribution-aware ensembles with clustering diagnostics offer robust, interpretable assessments of groundwater contamination.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper develops an ensemble ML framework for predicting the Heavy Metal Pollution Index (HPI) in Densu Basin groundwater. It compares raw, log, and Gaussian copula transformations of the skewed HPI target across six learners (SVM, k-NN, CART, Elastic Net, kernel ridge, and stacked Lasso), reporting that the copula + stacked ensemble yields the strongest results (R² = 0.96, RMSE = 0.19) with better residuals and spatially plausible maps. DBSCAN clustering identifies Fe and Mn as dominant contributors, aligning with regional hydrogeochemistry. The work notes reliance on random (non-spatial) nested CV as a limitation and calls for future spatial validation.

Significance. If the performance advantage survives spatial validation, the framework would supply a concrete, interpretable pipeline for handling non-Gaussian, correlated pollution indices in environmental monitoring. The explicit demonstration that raw-scale models produce near-perfect but over-optimistic fits, together with the clustering-hydrogeochemistry consistency check, strengthens the case for distribution-aware ensembles in groundwater studies.

major comments (2)
  1. [Methods and Results] Methods and Results sections (cross-validation description and performance tables): All reported metrics, including the headline Gaussian-copula stacked-ensemble result (R² = 0.96, RMSE = 0.19), were obtained with random nested cross-validation. Because groundwater samples exhibit spatial autocorrelation, random folds permit leakage between nearby points; the abstract flags this limitation yet supplies no quantitative check (spatial blocking, variogram of residuals, or comparison of random vs. spatial CV scores) to show that the claimed superiority of the copula transformation is robust to this bias.
  2. [Results] Results section (residual and map analysis): The claim that copula-based models produce 'improved residuals' and 'spatially plausible maps' is central to the reliability argument, but the manuscript provides neither quantitative residual diagnostics (e.g., Moran’s I, spatial variograms) nor side-by-side map comparisons with uncertainty bands. Without these, the visual and numerical improvement cannot be isolated from possible spatial leakage.
minor comments (2)
  1. [Abstract] Abstract: The list of learners ('six learners: SVM, k-NN, CART, Elastic Net, kernel ridge regression, and a stacked Lasso ensemble') is slightly ambiguous—clarify whether the stacked model counts as the sixth learner or is an additional meta-learner.
  2. [Clustering analysis] Clustering subsection: DBSCAN parameter choices (epsilon, min_samples) and any stability or validation metrics are not reported; adding these would strengthen the claim of consistency with hydrogeochemistry.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments, which highlight important considerations for spatial validation in environmental machine learning applications. We address each major comment below and outline targeted revisions to improve the manuscript's rigor.

read point-by-point responses

  1. Referee: [Methods and Results] Methods and Results sections (cross-validation description and performance tables): All reported metrics, including the headline Gaussian-copula stacked-ensemble result (R² = 0.96, RMSE = 0.19), were obtained with random nested cross-validation. Because groundwater samples exhibit spatial autocorrelation, random folds permit leakage between nearby points; the abstract flags this limitation yet supplies no quantitative check (spatial blocking, variogram of residuals, or comparison of random vs. spatial CV scores) to show that the claimed superiority of the copula transformation is robust to this bias.

    Authors: We agree that random nested cross-validation can introduce leakage due to spatial autocorrelation in groundwater data, potentially affecting the reliability of performance claims. Our current dataset size constrains the use of spatial blocking without compromising fold integrity, which is why we flagged this as a limitation and called for future spatial validation. In the revision, we will add a quantitative post-hoc check by computing Moran's I on the residuals of the Gaussian copula stacked ensemble and other top models to evaluate residual spatial structure. We will also expand the discussion to address how this impacts the relative advantage of the copula transformation. This provides the strongest feasible robustness assessment without requiring new data collection. revision: partial

  2. Referee: [Results] Results section (residual and map analysis): The claim that copula-based models produce 'improved residuals' and 'spatially plausible maps' is central to the reliability argument, but the manuscript provides neither quantitative residual diagnostics (e.g., Moran’s I, spatial variograms) nor side-by-side map comparisons with uncertainty bands. Without these, the visual and numerical improvement cannot be isolated from possible spatial leakage.

    Authors: We accept that additional quantitative diagnostics would strengthen the claims regarding residual improvement and map plausibility. In the revised manuscript, we will include Moran's I statistics and variogram analyses of residuals for the raw, log, and Gaussian copula transformations to provide objective measures of spatial structure. We will also add side-by-side predicted HPI maps with uncertainty bands (computed from ensemble variance) for direct comparison across transformations. These additions will help demonstrate that the reported improvements are not solely due to spatial leakage. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical ML performance on held-out data

full rationale

The paper reports standard empirical results from applying transformations (including Gaussian copula) and training ML models (SVM, k-NN, etc.) with nested cross-validation, then evaluating R²/RMSE on held-out folds. No mathematical derivation chain exists that reduces predictions to inputs by construction, no self-citations are load-bearing, and no fitted parameters are relabeled as independent predictions. The Gaussian copula is a conventional preprocessing step whose parameters are estimated from training data only; test metrics remain external to that fit. Spatial CV limitations are noted but do not create circularity in the reported numbers.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard machine learning assumptions and data transformations applied to environmental data; no new entities postulated. Free parameters are the fitted hyperparameters and transformation parameters.

free parameters (2)
  • Hyperparameters of SVM, k-NN, CART, Elastic Net, kernel ridge, and stacked Lasso
    Fitted via nested cross-validation for each transformation and learner.
  • Parameters of Gaussian copula transformation
    Estimated from data to model joint distributions of correlated contaminants.
axioms (2)
  • domain assumption Random nested cross-validation adequately captures predictive performance despite spatial heterogeneity in pollution data
    Invoked in evaluation but flagged as limitation in abstract.
  • domain assumption Transformations (raw, log, Gaussian copula) preserve predictive relationships without introducing bias
    Assumed when comparing model performance across transformations.

pith-pipeline@v0.9.0 · 5663 in / 1577 out tokens · 56016 ms · 2026-05-09T19:49:31.132839+00:00 · methodology

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

62 extracted references · 33 canonical work pages

  1. [1]

    In: 2021 International Conference on Electrical, Computer and Energy Technologies (ICECET), IEEE, pp 1--6

    Afrifa GY, Ansah-Narh T, Loh YSA, et al (2021) Estimation of groundwater heavy metal pollution indices via an amalgam of stack ensemble learning. In: 2021 International Conference on Electrical, Computer and Energy Technologies (ICECET), IEEE, pp 1--6

    2021

  2. [2]

    Turkish Journal of Emergency Medicine 18(3):91–93

    Akoglu H (2018) User’s guide to correlation coefficients. Turkish Journal of Emergency Medicine 18(3):91–93. doi:10.1016/j.tjem.2018.08.001

    work page doi:10.1016/j.tjem.2018.08.001 2018

  3. [3]

    Journal of Hydrology: Regional Studies 40:101,017

    Akurugu BA, Obuobie E, Yidana SM, et al (2022) Groundwater resources assessment in the densu basin: A review. Journal of Hydrology: Regional Studies 40:101,017

    2022

  4. [4]

    What about the other intervals? The American Statistician

    Altman NS (1992) An introduction to kernel and nearest-neighbor nonparametric regression. The American Statistician 46(3):175–185. doi:10.1080/00031305.1992.10475879

    work page doi:10.1080/00031305.1992.10475879 1992

  5. [5]

    Applied Water Science 1(1–2):41–48

    Amoako J, Karikari AY, Ansa-Asare OD (2011) Physico-chemical quality of boreholes in densu basin of ghana. Applied Water Science 1(1–2):41–48. doi:10.1007/s13201-011-0007-0

    work page doi:10.1007/s13201-011-0007-0 2011

  6. [6]

    International Scholarly Research Notices 2014:1–37

    Armah FA, Quansah R, Luginaah I (2014) A systematic review of heavy metals of anthropogenic origin in environmental media and biota in the context of gold mining in ghana. International Scholarly Research Notices 2014:1–37. doi:10.1155/2014/252148

    work page doi:10.1155/2014/252148 2014

  7. [7]

    Environmental Geology 36(1–2):55–64

    Backman B, Bodiš D, Lahermo P, et al (1998) Application of a groundwater contamination index in finland and slovakia. Environmental Geology 36(1–2):55–64. doi:10.1007/s002540050320, ://dx.doi.org/10.1007/s002540050320

    work page doi:10.1007/s002540050320 1998

  8. [8]

    Journal of the Royal Statistical Society Series B: Statistical Methodology 26(2):211--243

    Box GE, Cox DR (1964) An analysis of transformations. Journal of the Royal Statistical Society Series B: Statistical Methodology 26(2):211--243

    1964

  9. [9]

    wadsworth

    Brieman L, Friedman JH, Olshen RA, et al (1984) Classification and regression trees. wadsworth

    1984

  10. [10]

    Burnham and David R

    Burnham KP, Anderson DR (2004) Multimodel inference: Understanding aic and bic in model selection. Sociological Methods and Research 33(2):261–304. doi:10.1177/0049124104268644

    work page doi:10.1177/0049124104268644 2004

  11. [11]

    In: International workshop on multiple classifier systems, Springer, pp 1--15

    Dietterich TG (2000) Ensemble methods in machine learning. In: International workshop on multiple classifier systems, Springer, pp 1--15

    2000

  12. [12]

    Advances in neural information processing systems 9

    Drucker H, Burges CJ, Kaufman L, et al (1996) Support vector regression machines. Advances in neural information processing systems 9

    1996

  13. [13]

    Environmental Research Letters 7(2):021,003

    Edmunds WM (2012) Limits to the availability of groundwater in africa. Environmental Research Letters 7(2):021,003. doi:10.1088/1748-9326/7/2/021003

    work page doi:10.1088/1748-9326/7/2/021003 2012

  14. [14]

    Environmental Geochemistry and Health 46(10):409

    Eid MH, Awad M, Mohamed EA, et al (2024) Comprehensive approach integrating water quality index and toxic element analysis for environmental and health risk assessment enhanced by simulation techniques. Environmental Geochemistry and Health 46(10):409

    2024

  15. [15]

    International Journal of Environmental Research and Public Health 17(4):1245

    Eldaw E, Huang T, Elubid B, et al (2020) A novel approach for indexing heavy metals pollution to assess groundwater quality for drinking purposes. International Journal of Environmental Research and Public Health 17(4):1245

    2020

  16. [16]

    In: kdd, pp 226--231

    Ester M, Kriegel HP, Sander J, et al (1996) A density-based algorithm for discovering clusters in large spatial databases with noise. In: kdd, pp 226--231

    1996

  17. [17]

    Journal of hydrologic engineering 12(4):347--368

    Genest C, Favre AC (2007) Everything you always wanted to know about copula modeling but were afraid to ask. Journal of hydrologic engineering 12(4):347--368

    2007

  18. [18]

    Spatial Statistics 10:87--102

    Gr \"a ler B (2014) Modelling skewed spatial random fields through the spatial vine copula. Spatial Statistics 10:87--102

    2014

  19. [19]

    Journal of Hydrology 377(1–2):80–91

    Gupta HV, Kling H, Yilmaz KK, et al (2009) Decomposition of the mean squared error and nse performance criteria: Implications for improving hydrological modelling. Journal of Hydrology 377(1–2):80–91. doi:10.1016/j.jhydrol.2009.08.003

    work page doi:10.1016/j.jhydrol.2009.08.003 2009

  20. [20]

    The Elements of Statistical Learning

    Hastie T, Tibshirani R, Friedman J (2009) The Elements of Statistical Learning. Springer New York, doi:10.1007/978-0-387-84858-7

    work page doi:10.1007/978-0-387-84858-7 2009

  21. [21]

    PeerJ 6:e5518

    Hengl T, Nussbaum M, Wright MN, et al (2018) Random forest as a generic framework for predictive modeling of spatial and spatio-temporal variables. PeerJ 6:e5518. doi:10.7717/peerj.5518

    work page doi:10.7717/peerj.5518 2018

  22. [22]

    Computer Methods and Programs in Biomedicine 240:107,725

    Hutson AD, Yu H (2023) Exact inference around ordinal measures of association is often not exact. Computer Methods and Programs in Biomedicine 240:107,725. doi:10.1016/j.cmpb.2023.107725

    work page doi:10.1016/j.cmpb.2023.107725 2023

  23. [23]

    Pattern Recognition Letters 31(8):651--666

    Jain AK (2010) Data clustering: 50 years beyond k-means. Pattern Recognition Letters 31(8):651--666. doi:10.1016/j.patrec.2009.09.011

    work page doi:10.1016/j.patrec.2009.09.011 2010

  24. [24]

    Springer US, doi:10.1007/978-1-0716-1418-1

    James G, Witten D, Hastie T, et al (2021) An Introduction to Statistical Learning: with Applications in R. Springer US, doi:10.1007/978-1-0716-1418-1

    work page doi:10.1007/978-1-0716-1418-1 2021

  25. [25]

    Soft Computing 29(3):1331--1346

    Kazemi U, Soleimani S (2025) A new approach data processing: density-based spatial clustering of applications with noise (dbscan) clustering using game-theory. Soft Computing 29(3):1331--1346

    2025

  26. [26]

    Rockova and E

    Krupskii P, Huser R, Genton MG (2018) Factor copula models for replicated spatial data. Journal of the American Statistical Association 113(521):467–479. doi:10.1080/01621459.2016.1261712

    work page doi:10.1080/01621459.2016.1261712 2018

  27. [27]

    Springer New York, doi:10.1007/978-1-4614-6849-3

    Kuhn M, Johnson K (2013) Applied Predictive Modeling. Springer New York, doi:10.1007/978-1-4614-6849-3

    work page doi:10.1007/978-1-4614-6849-3 2013

  28. [28]

    Simon and Schuster

    Kunapuli G (2023) Ensemble methods for machine learning. Simon and Schuster

    2023

  29. [29]

    Applied geochemistry 17(5):569--581

    Lee G, Bigham JM, Faure G (2002) Removal of trace metals by coprecipitation with fe, al and mn from natural waters contaminated with acid mine drainage in the ducktown mining district, tennessee. Applied geochemistry 17(5):569--581

    2002

  30. [30]

    Guangxi Sci 25:393--399

    Li S, Xiong J, Deng C, et al (2018) The assessment of the heavy metal pollution and health risks in the liujiang river, xijiang region. Guangxi Sci 25:393--399

    2018

  31. [31]

    A Concordance Correlation Coefficient to Evaluate Reproducibility

    Lin LIK (1989) A concordance correlation coefficient to evaluate reproducibility. Biometrics 45(1):255. doi:10.2307/2532051

    work page doi:10.2307/2532051 1989

  32. [32]

    Environmental Research Letters 7(2):024,009

    MacDonald AM, Bonsor HC, Dochartaigh B \'E \'O , et al (2012) Quantitative maps of groundwater resources in africa. Environmental Research Letters 7(2):024,009

    2012

  33. [33]

    Environmental science and pollution research 23(8):7255--7265

    Mamat Z, Haximu S, Zhang ZY, et al (2016) An ecological risk assessment of heavy metal contamination in the surface sediments of bosten lake, northwest china. Environmental science and pollution research 23(8):7255--7265

    2016

  34. [34]

    Technology Innovation Office, Office of Solid Waste and Emergency Response

    McLean JE (1992) Behavior of metals in soils. Technology Innovation Office, Office of Solid Waste and Emergency Response

    1992

  35. [35]

    Journal of Environmental Science & Health Part A 31(2):283--289

    Mohan SV, Nithila P, Reddy SJ (1996) Estimation of heavy metals in drinking water and development of heavy metal pollution index. Journal of Environmental Science & Health Part A 31(2):283--289

    1996

  36. [36]

    Journal of Geographic Information System 8(05):618

    Nyamekye C, Nyame FK, Ofosu SA, et al (2016) Using geospatial information component to monitor the watersheds along the densu basin in ghana. Journal of Geographic Information System 8(05):618

    2016

  37. [37]

    Sustainable Chemistry for the Environment 2:100,015

    Osei-Owusu J, Heve WK, Duker RQ, et al (2023) Assessments of microbial and heavy metal contaminations in water supply systems at the university of environment and sustainable development in ghana. Sustainable Chemistry for the Environment 2:100,015. doi:10.1016/j.scenv.2023.100015

    work page doi:10.1016/j.scenv.2023.100015 2023

  38. [38]

    Environmental Geology 41(1–2):183–188

    Prasad B, Bose J (2001) Evaluation of the heavy metal pollution index for surface and spring water near a limestone mining area of the lower himalayas. Environmental Geology 41(1–2):183–188. doi:10.1007/s002540100380

    work page doi:10.1007/s002540100380 2001

  39. [39]

    Groundwater for Sustainable Development 9:100,245

    Rezaei A, Hassani H, Hassani S, et al (2019) Evaluation of groundwater quality and heavy metal pollution indices in bazman basin, southeastern iran. Groundwater for Sustainable Development 9:100,245. doi:10.1016/j.gsd.2019.100245

    work page doi:10.1016/j.gsd.2019.100245 2019

  40. [40]

    Ecography 40(8):913--929

    Roberts DR, Bahn V, Ciuti S, et al (2017) Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure. Ecography 40(8):913--929

    2017

  41. [41]

    Environmental Geochemistry and Health 45(12):9757–9784

    Saeed O, Székács A, Jordán G, et al (2023) Investigating the impacts of heavy metal(loid)s on ecology and human health in the lower basin of hungary’s danube river: A python and monte carlo simulation-based study. Environmental Geochemistry and Health 45(12):9757–9784. doi:10.1007/s10653-023-01769-4

    work page doi:10.1007/s10653-023-01769-4 2023

  42. [42]

    In: Proceedings of the 15th International Conference on Machine Learning (ICML 1998)

    Saunders C, Gammerman A, Vovk V (1998) Ridge regression learning algorithm in dual variables. In: Proceedings of the 15th International Conference on Machine Learning (ICML 1998). Morgan Kaufmann, San Francisco, CA, pp 515--521

    1998

  43. [43]

    Shmueli G (2010) To explain or to predict? Statistical science pp 289--310

    2010

  44. [44]

    Environment, Development and Sustainability 22(8):7847–7864

    Singh KR, Dutta R, Kalamdhad AS, et al (2019) Review of existing heavy metal contamination indices and development of an entropy-based improved indexing approach. Environment, Development and Sustainability 22(8):7847–7864. doi:10.1007/s10668-019-00549-4

    work page doi:10.1007/s10668-019-00549-4 2019

  45. [45]

    Applied Geochemistry 17(5):517--568

    Smedley PL, Kinniburgh DG (2002) A review of the source, behaviour and distribution of arsenic in natural waters. Applied Geochemistry 17(5):517--568. doi:10.1016/S0883-2927(02)00018-5

    work page doi:10.1016/s0883-2927(02)00018-5 2002

  46. [46]

    Statistics and Computing 14(3):199–222

    Smola AJ, Sch\" o lkopf B (2004) A tutorial on support vector regression. Statistics and Computing 14(3):199–222. doi:10.1023/b:stco.0000035301.49549.88

    work page doi:10.1023/b:stco.0000035301.49549.88 2004

  47. [47]

    Statistical Theory and Related Fields 6(1):87–87

    Sohil F, Sohali MU, Shabbir J (2021) An introduction to statistical learning with applications in r. Statistical Theory and Related Fields 6(1):87–87. doi:10.1080/24754269.2021.1980261

    work page doi:10.1080/24754269.2021.1980261 2021

  48. [48]

    In: Advances in pharmacology, vol 96

    Speer RM, Zhou X, Volk LB, et al (2023) Arsenic and cancer: Evidence and mechanisms. In: Advances in pharmacology, vol 96. Elsevier, p 151--202

    2023

  49. [49]

    West African Journal of Applied Ecology 12(1)

    Tay C, Kortatsi B (2008) Groundwater quality studies: A case study of the densu basin, ghana. West African Journal of Applied Ecology 12(1). doi:10.4314/wajae.v12i1.45760

    work page doi:10.4314/wajae.v12i1.45760 2008

  50. [50]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , author =

    Tibshirani R (1996) Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society Series B: Statistical Methodology 58(1):267–288. doi:10.1111/j.2517-6161.1996.tb02080.x

    work page doi:10.1111/j.2517-6161.1996.tb02080.x 1996

  51. [51]

    Journal of Artificial Intelligence Research 10:271–289

    Ting KM, Witten IH (1999) Issues in stacked generalization. Journal of Artificial Intelligence Research 10:271–289. doi:10.1613/jair.594

    work page doi:10.1613/jair.594 1999

  52. [52]

    Bias in error estimation when using cross- validation for model selection,

    Varma S, Simon R (2006) Bias in error estimation when using cross-validation for model selection. BMC Bioinformatics 7(1). doi:10.1186/1471-2105-7-91

    work page doi:10.1186/1471-2105-7-91 2006

  53. [53]

    World Health Organization

    WHO (2022) Guidelines for drinking-water quality: incorporating the first and second addenda. World Health Organization

    2022

  54. [54]

    Climate Research 30:79–82

    Willmott C, Matsuura K (2005) Advantages of the mean absolute error (mae) over the root mean square error (rmse) in assessing average model performance. Climate Research 30:79–82. doi:10.3354/cr030079

    work page doi:10.3354/cr030079 2005

  55. [55]

    In: Neural Networks

    Wolpert DH (1992) Stacked generalization. Neural Networks 5(2):241–259. doi:10.1016/s0893-6080(05)80023-1

    work page doi:10.1016/s0893-6080(05)80023-1 1992

  56. [56]

    Communications in Statistics - Theory and Methods 53(6):2141–2153

    Yu H, Hutson AD (2022) A robust spearman correlation coefficient permutation test. Communications in Statistics - Theory and Methods 53(6):2141–2153. doi:10.1080/03610926.2022.2121144

    work page doi:10.1080/03610926.2022.2121144 2022

  57. [57]

    International Journal of Environmental Research and Public Health 19(6):3571

    Zhai Y, Zheng F, Li D, et al (2022) Distribution, genesis, and human health risks of groundwater heavy metals impacted by the typical setting of songnen plain of ne china. International Journal of Environmental Research and Public Health 19(6):3571. doi:10.3390/ijerph19063571

    work page doi:10.3390/ijerph19063571 2022

  58. [58]

    In: Proceedings of the 14th ACM international conference on web search and data mining, pp 418--426

    Zhang Z, Rudra K, Anand A (2021) Explain and predict, and then predict again. In: Proceedings of the 14th ACM international conference on web search and data mining, pp 418--426

    2021

  59. [59]

    CRC press

    Zhou ZH (2025) Ensemble methods: foundations and algorithms. CRC press

    2025

  60. [60]

    Science of the total environment 275(1-3):19--26

    Zietz B, de Vergara JD, Kevekordes S, et al (2001) Lead contamination in tap water of households with children in lower saxony, germany. Science of the total environment 275(1-3):19--26

    2001

  61. [61]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , author =

    Zou H, Hastie T (2005) Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society Series B: Statistical Methodology 67(2):301–320. doi:10.1111/j.1467-9868.2005.00503.x

    work page doi:10.1111/j.1467-9868.2005.00503.x 2005

  62. [62]

    Geochemistry: Exploration, Environment, Analysis 19(2):129--137

    Z \'u \ n iga-V \'a zquez D, Armienta MA, Deng Y, et al (2019) Evaluation of fe, zn, pb, cd and as mobility from tailings by sequential extraction and experiments under imposed physico-chemical conditions. Geochemistry: Exploration, Environment, Analysis 19(2):129--137

    2019