• No results found

Advances in multidimensional unfolding Busing, F.M.T.A.

N/A
N/A
Protected

Academic year: 2021

Share "Advances in multidimensional unfolding Busing, F.M.T.A."

Copied!
3
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Advances in multidimensional unfolding

Busing, F.M.T.A.

Citation

Busing, F. M. T. A. (2010, April 21). Advances in multidimensional unfolding. Retrieved from https://hdl.handle.net/1887/15279

Version: Not Applicable (or Unknown)

License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden

Downloaded from: https://hdl.handle.net/1887/15279

Note: To cite this publication please use the final published version (if

applicable).

(2)

glossary of solutions

This glossary contains terms directly related with degenerate solutions, which, in general, are zero stress solutions that fail to preserve the structure that is contained in the data. Other terms can be found through the subject index.

Trivial solution — A perfect but meaningless solution having zero stress.

Often, this solution can be avoided by using a proper normalization of the stress function. The most renowned trivial solutions are the two-point or object-sphere solution for unconditional unfolding using raw stress, the two-plus-two-point solution using stress-

1

, and the object-point solution for row-conditional unfolding using stress-

2

with an improper summation for the normalization factor.

Absolutely or completely degenerate solution — A flawless but purposeless zero stress solution without variation in the transformed preferences or the distances that cannot be avoided by using a proper normalization of the stress function. An absolutely degenerate solution resembles a trivial solution, such as the object-sphere solution, but differs in computational respect to that solution.

Partially degenerate solution — An impeccable but ineffectual row-conditional zero stress solution without variation in the rows of the transformed prefer- ences or the distances that cannot be avoided by using a proper normalization of the stress function. A partially degenerate solution resembles a trivial solution, for example an object-point solution, but deviates in view of compu- tational aspects.

Extremely, near(ly), quasi, strongly, or approximately degenerate solution — A solution that tends to a absolutely or partially degenerated solution, but deviates from this solution due to non-convergence, local minima, or other anomalies. Such a solution is difficult to identify, because only part of the solution is absolutely or partially degenerate. Commonly, this solution shows conspicuously low variation in either the transformed preferences or the distances and has a strikingly low stress value.

(3)

glossary of solutions

For the following solutions, subjects (rows) are represented by a small dot (one subject) or a large dot (multiple subject) and objects (columns) are represented by a plus sign (one object) or a cross (multiple objects).

Two-point solution — This zero stress solution holds only one distance value for all subject-object distances.

This equal distance solution is the result of minimiz- ing raw stress without a normalization factor. The transformation of the preferences results in one value for all transformed preferences, i.e., γ = c, and with distances identical to this value, i.e., d = c, r-stress kγ − dk2is equal to zero.

Two-plus-two-point solution — This zero stress so- lution contains one distance value for all subject- object distances, except for one subject-object dis- tance, which has a different value. This solution is the result of minimizing stress-

1

with the sum-of- squares of the distances as normalization factor. Due to the one differing distance value, and correspond- ing transformed preferences, the normalization fac- tor kdk2 remains unequal to zero, while stress-

1

kγ − dk2/kdk2is equal to zero.

Object-sphere solution — This zero stress solution forms an extension of the two-point solution as there only exist one value for the distances. Whether the subjects are on the edge of the circle with the object in the center or the objects are on the fringe with the subjects in the middle, is of subordinate significance, since only subject-object distances matter for the loss function.

Object-point solution — This zero stress solution con- cerns a specific row-conditional solution with an im- proper normalization factor. Each subject has one value for the subject-object distances, but different values may exist for different subjects. A normaliza- tion factor, even a variance-based normalization fac- tor as used in stress-

2

or s-stress-

2

, that considers all distances simultaneously (i.e., improper summa- tion), remains unequal to zero, while the raw stress part of the loss function is equal to zero.

224

Referenties

GERELATEERDE DOCUMENTEN

genfold- 2 (DeSarbo & Rao, 1984) (Kim, Rangaswamy, & DeSarbo, 1999, even mention a genfold) was never made publicly available, but the loss function is simple enough to

In this chapter, a simple but effective solution is proposed to the degener- acy problem in metric unfolding: Penalize the undesirable intercept to avoid identical

To avoid such degeneracies, Kruskal and Carroll (1969) proposed alterna- tive badness-of-fit functions that incorporate normalization factors based on the variance of the

This rules in favor of internal preference analysis (over external preference analysis) and we will therefor compare the restricted unfolding solution with three such analyses, a

Figure 6.7 Boxplots of the recovery measures ( φ xy , φ y , and τ b , for the left-hand, middle, and right-hand panels, respectively; lower left-hand panel includes both φ high

Examples of such devel- opments were presented in subsequent chapters: Chapter 5 elaborated on a previously published model extension, restricting the coordinates to be

Depending on the individual differences model and the restrictions on the configuration, updates are determined for the row coordinates X, the column coordinates Y, the space weights

Relating preference data to multidimensional scaling solutions via a generalization of Coombs’ unfolding model.. Bell