A Comprehensive Workflow for Compositional Data Analysis in Archaeometry, with Code in R DOI Creative Commons
Michael Greenacre, Jonathan R. Wood

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: Aug. 2, 2024

Abstract Compositional data, which have relative rather than absolute meaning, are common in quantitative archaeological research. Such multivariate data usually expressed as proportions, summing to 1, or equivalently percentages. We present a comprehensive and defensible workflow for processing compositional archaeometry, using both the original values their transformation logratios. The most useful logratio transformations illustrated how they affect interpretation of final results context unsupervised supervised learning. is on from bronze ritual vessels provide fingerprints Shang Zhou periods Chinese Bronze Age. Predictions, with caveats, fabrication age made -- effect, typological seriation bronzes. In Supplementary Material, we further explore effect zeros dataset compare logaratio analyses chiPower approach, where replace any value determined being below detection limit instruments element, zeros. R code reproducing all provided Material online.

Language: Английский

Safety assurance and nutritional quality enhancement of Phyllospora comosa biomass using hydrothermal treatments derived ensemble machine learning models DOI

Thiru Chenduran Somasundaram,

Thomas S. Mock, Damien L. Callahan

et al.

SSRN Electronic Journal, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 1, 2024

One of the major barriers to mass industrial utilization brown seaweeds as foodsources stem from food safety risks associated with their iodine and arsenicconcentrations, which typically exceed regulatory limits. Hydrothermal treatments mighteffectively reduce arsenic concentration Phyllospora comosa belowAustralian maximum residual limits (iodine = 1 mg/g 0.00667 mg/g; dryweight). The experimental hydrothermal dictated that 82 °C – 250 streatment reduced 2.76 0.88 (68%reduction) 0.01693 0.00965 (43%reduction). Machine learning models predicted blanching at 100 for ~ 4 minuteswill below its limit. This researchshows hydrothermally treated products are safer humanconsumption thus may permit expansion seaweed product production andconsumption in Australia.

Language: Английский

Citations

3
The chiPower transformation: a valid alternative to logratio transformations in compositional data analysis DOI
Michael Greenacre

Advances in Data Analysis and Classification, Journal Year: 2024, Volume and Issue: 18(3), P. 769 - 796

Published: Aug. 1, 2024

Language: Английский

Citations

3
Diffuse reflectance mid infra-red spectroscopy combined with machine learning algorithms can differentiate spectral signatures in shallow and deeper soils for the prediction of pH and organic matter content DOI Creative Commons
Felipe Bachion de Santana, Eric Grunsky, Mairéad M. Fitzsimons

et al.

CATENA, Journal Year: 2022, Volume and Issue: 218, P. 106552 - 106552

Published: Aug. 3, 2022

Precision and sustainable agriculture requires information about soil pH organic matter (OM) content at higher spatial temporal scales than current agronomic sampling analytical methods allow. This study examined the accuracy of spectral models using high throughput screening (HTS) in diffuse reflectance mode mid Infra-red (MIR)/DRIFT combined with machine learning algorithms to predict pH(CaCl2) %OM shallow deeper topsoils compared laboratory methods. Models were developed from an archive samples taken on a 4 km2 grid northern half Ireland (Terra Soil project), which includes 18,859 (9,396 + 9,463 deeper). The application Cubist showed that for different depths there are minor group associations values. These differences resulted loss extrapolation topsoil model values or vice versa. Therefore we recommend use both build calibration model. proposed methodology was able determine unique multivariate regression depths, RMSEP 1.12 0.89 %; RPIQ 42.34 38.48; R2val 0.9989 0.9993 determinations topsoils, respectively. For obtained 0.25 0.34; 6.04 4.94; 0.9385 0.8954. Both classified as excellent predictions models, yielding >4.05 topsoils. results demonstrated potential HTS-DRIFT rapid, accurate, cost-effective method large libraries, displaying predicted similar two separate (pH LOI).

Language: Английский

Citations

13
Identification of spatial clusters of potentially toxic elements in different soil types using unsupervised machine learning and compositional data analysis DOI Creative Commons
Gevorg Tepanosyan, Zhenya Poghosyan, Lilit Sahakyan

et al.

Soil & Environmental Health, Journal Year: 2024, Volume and Issue: 2(3), P. 100085 - 100085

Published: May 8, 2024

Soil health is an important concept and the generation of soil chemical composition data, including potentially toxic elements (PTEs) one its conceptual components. This study aims to reveal spatial distribution patterns PTEs contents, identify their potential sources, unveil geochemical associations in Aragatsotn region (Armenia). For that purpose, contents Cr, V, Ti, As, Zn, Cu, Co, Fe, Mn, Ba, Pb were determined using X-ray fluorescence spectrometer (Innov X-5000). The results showed mean Cr As exceeded upper continental crust by 1.50 3.12 times maximum acceptable values by1.53, times, respectively. detailed analysis pattern demonstrated presence sites where all these displayed comparatively higher values. combined application compositional data geospatial mapping allowed multivariate outliers which located structural-metallogenic zones with active exploitation ore deposits. unsupervised machine learning algorithm unveiled three groups within main dataset clr-biplot identified source-specific elements. Particularly, Group I included Cu highest value among groups. samples found areas Calcisols developed high attributed agricultural activities (vineyards) vehicle emissions. II represented association As. formation this group conditioned volcanic rocks local geological origin. However, no was aligned types. III Pb, Ba. may have a mixed origin as it spatially distributed regional highways pass through also exhibit

Language: Английский

Citations

2
Power transformations of relative count data as a shrinkage problem DOI
Ionas Erb

Information Geometry, Journal Year: 2023, Volume and Issue: 6(1), P. 327 - 354

Published: April 13, 2023

Language: Английский

Citations

4
A Log‐Ratio‐Based Algorithm for Petrologic Mass‐Balance Problems and Uncertainty Assessment DOI Creative Commons
Kelsey B. Prissel, Jean‐Arthur Olive, M. J. Krawczynski

et al.

Geochemistry Geophysics Geosystems, Journal Year: 2023, Volume and Issue: 24(12)

Published: Dec. 1, 2023

Abstract We provide a new algorithm for mass‐balance calculations in petrology and geochemistry based on the log‐ratio approach championed initially by John Aitchison (e.g., Aitchison, 1982, https://doi.org/10.1111/j.2517-6161.1982.tb01195.x ; 1984, https://doi.org/10.1007/bf01029316 ) along with underlying principles, mathematical frameworks, data requirements. Log‐ratio Inversion of Mixed End‐members (LIME) is written MATLAB calculates phase proportions an experiment or rock given bulk composition, composition each phase, associated compositional uncertainties. An important advantage LIME that performing calculation inverse space constrains to be between 0 100 wt.%. Further, resulting realistic estimates uncertainty regardless distribution. These two characteristics improve upon standard multiple linear regression techniques, which may yield negative values if non‐constrained report oversimplified symmetric errors. Primary applications include estimating abundances, calculating melting metamorphic reaction stoichiometries, checking open system behavior equilibria experiments. The technique presented here covers whole‐rock analysis, mineralogy, abundance, but could extended isotopic tracers, trace element modeling, regolith component un‐mixing. highlight importance estimations abundances fields comparing our results from previously published calculations. Furthermore, we present case studies demonstrate role determining magma crystallinity defining reactions.

Language: Английский

Citations

4
Compositional-geochemical characterization of lead (Pb) anomalies and Pb-induced human health risk in urban topsoil DOI

Gevorg Tepanosyan,

Astghik Gevorgyan, Stefano Albanese

et al.

Environmental Geochemistry and Health, Journal Year: 2024, Volume and Issue: 46(6)

Published: May 2, 2024

Language: Английский

Citations

1
The brute force simulation of the nutrient losses during aquaculture feed hydrothermal processing using Phyllospora comosa compositional data DOI

Thiru Chenduran Somasundaram,

Thomas S. Mock, Damien L. Callahan

et al.

SSRN Electronic Journal, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 1, 2024

Nutrient compositional contrasts of processed and unprocessed feed samples concealtrue behaviour nutrients during processing experiments. A reference nutrientbased additive log ratio transformation alleviates these constraints. Areference nutrient is selected based on its process stability. This paper describes twoobjective methods for the estimation hydrothermal stability Phyllosporacomosa proximate nutrients. The unit proportion method calculates total mass lossrequired by each to reach final composition. brute force simulationestimates true retentions at their minimal loss. lipidcontent resulted as most stable in both methods. lipid additivelog-ratio analysis large, significant (P <0.05), andnegative effects treatments contents. describedherein can be applied across various selection ofbiomarkers using contentual changes behaviours

Language: Английский

Citations

1
A simple approach to dealing with partial contestation DOI Creative Commons
Ali Kagalwala, Thiago Moreira, Guy D. Whitten

et al.

Social Science Quarterly, Journal Year: 2024, Volume and Issue: 105(4), P. 948 - 964

Published: July 1, 2024

Abstract Objective We propose a simple approach to dealing with partial contestation in models of multiparty elections. Methods Our proposed is add tiny value the vote share parties that do not contest district and then include dummy variables identifying those districts which compete. can estimate single system equations using seemingly unrelated regression (SUR) Aitchison's log‐ratio transformation. In our SUR system, we interact variable for party partially contested other predictors equation uses votes same outcome. Finally, robust standard errors this address heteroscedasticity. Results demonstrate utility simulated data election results from English parliamentary elections 2017. Conclusion From simulations, find recommended performs as well by Tom, Tucker, Wittenberg. strategy advantageous it easy estimate, information all districts, addresses real‐world regressions.

Language: Английский

Citations

1
Hierarchical mixture of discriminative Generalized Dirichlet classifiers DOI Creative Commons
Elvis Togban, Djemel Ziou

Pattern Recognition, Journal Year: 2024, Volume and Issue: 156, P. 110789 - 110789

Published: July 15, 2024

This paper presents a discriminative classifier for compositional data. is based on the posterior distribution of Generalized Dirichlet which counterpart mixture model. Moreover, following experts paradigm, we proposed hierarchical this classifier. In order to learn models parameters, use variational approximation by deriving an upper-bound mixture. To best our knownledge, first time bound in literature. Experimental results are presented spam detection and color space identification.

Language: Английский

Citations

1