Alternating coupled vectors resolution (ACOVER) method for trilinear analysis of three-way data

✍ Scribed by Jian-Hui Jiang; Hai-Long Wu; Yang Li; Ru-Qin Yu

• Publisher: John Wiley and Sons
• Year: 1999
• Tongue: English
• Weight: 163 KB
• Volume: 13
• Category: Article
• ISSN: 0886-9383
• DOI: 10.1002/(sici)1099-128x(199911/12)13:6<557::aid-cem563>3.0.co;2-c

✦ Synopsis

A new method, alternating coupled vectors resolution (ACOVER), is developed for trilinear analysis of a threeway data array. First, based on the least-squares principle, four coupled vectors resolution (COVER) errors are proposed. Second, a procedure which resolves the profiles of each component successively is developed. This characteristic of successive resolution provides a natural way to avoid the two-factor degeneracy. Experimental results show that the ACOVER method has the advantage that the resolved profiles of analytical interest are very stable with respect to the estimated component number when the number of components is chosen to be equal to or greater than the actual model dimensionality. This circumvents the dilemma of determining a proper component number for the model, which is difficult to handle for the PARAFAC algorithm. Moreover, the method has much higher convergence rate than the PARAFAC algorithm.