Journal of Physical and Chemical Reference Data · 2022 · 11 citations · 228 references
Natural SciencesChemical EngineeringChemistryAnalytical ChemistryElectrochemistryInorganic ChemistryDensity EstimationElectrolyte SolutionsAqueous Electrolyte SolutionsBenchmark DatasetsData DistributionData EvaluationGaussian Mixture ModelsReference ElectrolytesSupport Vector RegressionSupport Vector Regression Surrogates
A reference density database of aqueous alkali halide solutions is presented. The solutes are the 20 alkali halides consisting of the cations Li+, Na+, K+, Rb+, Cs+, and anions F−, Cl−, Br−, I−. Experimental density data of these aqueous electrolyte solutions are extensively collected and critically evaluated. A data evaluation procedure is proposed, utilizing support vector regression (SVR) to compare the experimental datasets against each other. Data evaluation is based on agreement with data from other sources rather than accuracy claimed in the literature. Datasets with large deviation from others are identified and removed. The proposed reference database consists of 11 081 data points of 586 datasets from 309 references. Maximum deviations between the selected datasets do not exceed 1%, and are smaller than 0.5% for most of the aqueous alkali halide solutions. SVR models are also trained based on the reference database. Data distribution is visualized using a Gaussian mixture model. Applicability domains of the SVR models are analyzed using Williams plots. An executable program is provided for calculations based on the SVR models.
228
Estimating the Support of a High-Dimensional Distribution
Bernhard Schölkopf, John Platt, John Shawe‐Taylor et al. · Neural Computation · 2001 · 5.9K citations
The expectation-maximization algorithm
Todd K. Moon · IEEE Signal Processing Magazine · 1996 · 3.2K citations
Jacques E. Desnoyers, Marcel Arel, Gérald Perron et al. · The Journal of Physical Chemistry · 1969 · 341 citations