Talk:Multidimensional scaling/Archives/2019

Unreadable

This article is unreadable. Look at the section "classical multidimensional scaling." What is $ X $? What is a coordinate matrix? What is $ 1 $? Is $ n $ the same as $ N $? This is throughout the article, never-mind how weird the overall presentation is. 217.92.118.242 (talk) 17:10, 5 May 2018 (UTC)

I agree with the preceding comment. I have a basic knowledge of PCA and have used it in research projects in the past - but I cannot make head nor tail of this article. Even the image at the top of the page is confusing. The caption claims that each red dot represents one Republican member and each blue dot represents one Democratic member. However, I think it should say that each dot represents a vote for a given senator or a the demographic/ psycho-social characteristics of a voter - after all the title reads "Voting patterns" which implies that it is votes, rather than politicians, which are the focus of the analysis. Just logically, it doesn't make much sense to sort politicians into groups (they are already sorted into the most salient dimension i.e., political affilication). From a statistical viewpoint, it is also questionable as to whether there would be a sufficient number of politicians to support multi-dimensional scaling which typically requires large sample sizes. Conversely, it makes much more sense to sort voting intentions or voters into groups, according to selected relevant criteria.

Too many Wikipedia articles go to great lengths to make their subject matter sound more complex - as if complexity somehow equates with importance and this article is a classic example of that approach. This is a common misunderstanding. The hallmark of good writing is to take complex ideas and explain them simply and clearly. BronHiggs (talk) 03:16, 25 August 2018 (UTC)

I agree that the lead especially needs to give clear information rather than jargon. I found a much better description which I hope will help: "The set of procedures referred to as multidimensional scaling methods are concerned with constructing a configuration of n points, usually in Euclidean space, from information about the pairwise 'distances' among a set of n objects or individuals." Anybody else with clear information about this topic, please add some more references we can use to improve the article. HouseOfChange (talk) 22:54, 14 February 2019 (UTC)

References