Diagnostics in a simple correspondence analysis model: An approach based on Cook’s distance for log-linear models

Martín, Nirian

doi:10.1016/j.jmva.2015.01.008

Cited by 3 publications

(1 citation statement)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Accordingly, Figure S6a diagnosed a random distribution without any patterns or trends, while the actual and predicted data in Figure S6b had high-compatibility because these points were randomly scattered along the 45-degree line. Also, Cook's distance in Figure S7a was used to record any changes made by the quadratic model in the case of deleting or omitting any data point [52]. Although the Cook's distance of two points was found to be larger than 1.0, which may lead to a lack of their accuracy, the others (18 data points) were lower than 1.0, meaning that omitting one of them is highly unlikely to vary the estimate of the regression coefficients [53].…”

Section: Optimization Studymentioning

confidence: 99%

Combined Minimum-Run Resolution IV and Central Composite Design for Optimized Removal of the Tetracycline Drug Over Metal–Organic Framework-Templated Porous Carbon

Tran

Nguyen

Le³

et al. 2019

Molecules

View full text Add to dashboard Cite

In this study, a minimum-run resolution IV and central composite design have been developed to optimize tetracycline removal efficiency over mesoporous carbon derived from the metal-organic framework MIL-53 (Fe) as a self-sacrificial template. Firstly, minimum-run resolution IV, powered by the Design–Expert program, was used as an efficient and reliable screening study for investigating a set of seven factors, these were: tetracycline concentration (A: 5–15 mg/g), dose of mesoporous carbons (MPC) (B: 0.05–0.15 g/L), initial pH level (C: 2–10), contact time (D: 1–3 h), temperature (E: 20–40 °C), shaking speed (F: 150–250 rpm), and Na+ ionic strength (G: 10–90 mM) at both low (−1) and high (+1) levels, for investigation of the data ranges. The 20-trial model was analyzed and assessed by Analysis of Variance (ANOVA) data, and diagnostic plots (e.g., the Pareto chart, and half-normal and normal probability plots). Based on minimum-run resolution IV, three factors, including tetracycline concentration (A), dose of MPC (B), and initial pH (C), were selected to carry out the optimization study using a central composite design. The proposed quadratic model was found to be statistically significant at the 95% confidence level due to a low P-value (<0.05), high R2 (0.9078), and the AP ratio (11.4), along with an abundance of diagnostic plots (3D response surfaces, Cook’s distance, Box-Cox, DFFITS, Leverage versus run, residuals versus runs, and actual versus predicted). Under response surface methodology-optimized conditions (e.g., tetracycline concentration of 1.9 mg/g, MPC dose of 0.15 g/L, and pH level of 3.9), the highest tetracycline removal efficiency via confirmation tests reached up to 98.0%–99.7%. Also, kinetic intraparticle diffusion and isotherm models were systematically studied to interpret how tetracycline molecules were absorbed on an MPC structure. In particular, the adsorption mechanisms including “electrostatic attraction” and “π–π interaction” were proposed.

show abstract

Section: Optimization Studymentioning

confidence: 99%

Combined Minimum-Run Resolution IV and Central Composite Design for Optimized Removal of the Tetracycline Drug Over Metal–Organic Framework-Templated Porous Carbon

Tran

Nguyen

Le³

et al. 2019

Molecules

View full text Add to dashboard Cite

show abstract

Microwave-assisted solvothermal fabrication of hybrid zeolitic–imidazolate framework (ZIF-8) for optimizing dyes adsorption efficiency using response surface methodology

Tran

Nguyen

Pham

et al. 2020

Journal of Environmental Chemical Engineering

View full text Add to dashboard Cite

Contribution of a categorical statistical test in examining dependencies among qualitative variables by means simplification of correspondence analysis

Ginanjar

Nurhuda

Sunengsih

et al. 2019

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

The contribution statistical test of categories to dependencies between qualitative variables is needed, because the existence of these categories will influence decision making. Decisions taken can be misleading, if a category does not contribute. If N is the matrix from 2×J or I×2 contingency table which is constructed from two qualitative variables, then the principal coordinates can be calculated using the simplification of correspondence analysis (SoCA). Principal coordinates produced by the SoCA are the same as conventional correspondence analysis (CA), but the SoCA presents a simpler calculation, because it is calculated directly from the elements of N. Principal coordinates derived from 2×J matrix N can be used to test the contribution of categories to dependencies between qualitative variables more simply, and obtained confidence intervals for each category.

show abstract

Diagnostics in a simple correspondence analysis model: An approach based on Cook’s distance for log-linear models

Cited by 3 publications

References 24 publications

Combined Minimum-Run Resolution IV and Central Composite Design for Optimized Removal of the Tetracycline Drug Over Metal–Organic Framework-Templated Porous Carbon

Combined Minimum-Run Resolution IV and Central Composite Design for Optimized Removal of the Tetracycline Drug Over Metal–Organic Framework-Templated Porous Carbon

Microwave-assisted solvothermal fabrication of hybrid zeolitic–imidazolate framework (ZIF-8) for optimizing dyes adsorption efficiency using response surface methodology

Contribution of a categorical statistical test in examining dependencies among qualitative variables by means simplification of correspondence analysis

Contact Info

Product

Resources

About