Multicriteria decision making: A case study

Keller, H.R.; Massart, D.L.; Brans, Jean Pierre

doi:10.1016/0169-7439(91)80064-w

Cited by 136 publications

(87 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…PCA is a well known chemometrics method [40,41] facilitating pattern recognition and display of data. PCA effects multi-variate data reduction by transforming the data into orthogonal components, which are linear combinations of the original variables.…”

Section: Principal Component Analysis (Pca)mentioning

confidence: 99%

“…PROMETHEE and GAIA MCDM methods have been described in detail elsewhere [40,43,44]; in this paper, a summary of these methods is provided in sufficient detail to indicate the concepts and data treatment involved.…”

Section: Promethee and Gaiamentioning

confidence: 99%

“…Decision Lab (Visual Decision 2000)), six shapes of preference functions are defined. In this software, they are called Usual, Linear, Level, Vshaped, U-shaped, and Gaussian [40]. For most functions, one or two classification thresholds must be provided by the user.…”

Section: Promethee and Gaiamentioning

confidence: 99%

“…5. The outranking flows are compared pairwise according to a set of rules [40]. These are based on three possible outcomes, which lead to partial pre-order of the objects known as PROMETHEE I ranking: a) one action is preferred to another; b) there is no difference between the two actions; and c) the two actions can not be compared.…”

Section: Promethee and Gaiamentioning

confidence: 99%

See 3 more Smart Citations

A spectroscopic study of mechanochemically activated kaolinite with the aid of chemometrics

Carmody

Kristóf

Frost

et al. 2005

Journal of Colloid and Interface Science

View full text Add to dashboard Cite

The study of kaolinite surfaces is of industrial importance. In this work we report the application of chemometrics to the study of modified kaolinite surfaces. DRIFT spectra of mechanochemically activated kaolinites (Kiralhegy, Zettlitz, Szegi and Birdwood) were analysed using, Principal Component Analysis (PCA), and multi-criteria decision making (MCDM) methods, PROMETHEE and GAIA. The clear discrimination of the Kiralhegy spectral objects on the two PC scores plots (400-800 and 800-2030 cm -1 ) indicated the dominance of quartz. Importantly, no ordering of any spectral objects appeared to be related to grinding time in the PC plots of these spectral regions. Thus, neither the kaolinite nor the quartz, are systematically responsive to grinding time according to the spectral criteria investigated. The third spectral region (2600-3800 cm -1 -OH vibrations), showed apparent systematic ordering of the Kiralhegy and, to a lesser extent, Zettlitz spectral objects with grinding time. This was attributed to the effect of the natural quartz on the delamination of kaolinite and the accompanying phenomena (i.e. formation of kaolinite spheres and water). With the MCDM methods, it was shown that useful information on the basis of chemical composition, physical properties and grinding time can be obtained. For example, the effects of the minor chemical components (e.g. MgO, K 2 O etc) indicated that the Birdwood kaolinite is arguably the most pure one analysed. In another MCDM experiment, some support was obtained for the apparent trend with grinding time noted in the PC plot of the OH spectral region.

show abstract

Section: Principal Component Analysis (Pca)mentioning

confidence: 99%

Section: Promethee and Gaiamentioning

confidence: 99%

Section: Promethee and Gaiamentioning

confidence: 99%

Section: Promethee and Gaiamentioning

confidence: 99%

See 2 more Smart Citations

A spectroscopic study of mechanochemically activated kaolinite with the aid of chemometrics

Carmody

Kristóf

Frost

et al. 2005

Journal of Colloid and Interface Science

View full text Add to dashboard Cite

show abstract

“…Iako ne postoji ograničenje u pogledu broja odgovora koji se na ovaj način mogu istovremeno optimizirati, što je njihov broj veći, manje su šanse u iznalaženju optimalne oblasti u okviru koje su svi razmatrani odgovori sistema prihvatljivi. Kod metoda Pareto optimizacije, eksperiment se smatra Pareto optimalnim ukoliko ne postoji ni jedan eksperiment sa boljim ishodom za jedan odgovor sistema, a da se pri tome ne dobije lošiji rezultat za drugi posmatrani odgovor sistema [32,33]. Na slici 6 prikazani su podaci za dva odgovora sistema (y 1 = vreme trajanja analize i y 2 = faktor rezolucije) dobijeni izvođenjem devet eksperimenata definisanih centralnim kompozicionim dizajnom za dva faktora.…”

Section: Slikaunclassified