|WikiProject Statistics||(Rated Stub-class, Low-importance)|
|This is the talk page for discussing improvements to the Ranking article.|
Hopefully cleaned up this page significantly. There may be some useful info in the German article. Removed the link to the French article which was about Rank (mathematics). Matthewmayer 02:03, 12 December 2005 (UTC)
I've expanded the summary a bit, and added a whole section on ranking strategies. I'm painfully aware this lacks any references, and would welcome anyone's efforts to rectify that. I'll try and come back to it myself when I have time to hunt down the necessary sources. Mooncow 00:46, 11 October 2006 (UTC)
Rank tests in statistics
- Started. DFH 08:23, 23 January 2007 (UTC)
The quality of Excel (or the lack thereof) should be discussed under the topic "Excel". Rather than giving a reference to the macro by Pottel, midranks can be computed within Excel by computing u-scores as RANK(A1,A$1:A$3,1)-RANK(A1,A$1:A$3,0) and then transforming them into ranks by adding COUNT(A$1:A$3)+1 and dividing by 2.
Since I don't want to subject myself to the same criticism, I would like to invite somebody else to have a look at mu-scores (u-scores for multivariate data), e.g., in (Wittkowski 2004, Statistics in Medicine), with a recent application to Tour de France data in (Cherchye Vermeulen, 2006, Journal of Sports Economics) —The preceding unsigned comment was added by Kmwittko (talk • contribs) 18:19:02, August 19, 2007 (UTC).
Should there be a citation for the various types of rankings that are listed in this article? Is the term 'Dense Rank' actually defined somewhere authoritatively? 184.108.40.206 (talk) 18:58, 1 April 2008 (UTC)eeverman
Strategies for Ranking
This section is a misnomer. It doesn't offer any useful insight whatsoever into how ranking is done, i.e., what actual strategies are employed. All it does is offer a fairly lame list of examples of what the resulting rank value assignments might look like, e.g., whether there might be duplicates or missing values.
It would be far more helpful to distinguish between situations where direct comparisons can be employed and those situations where that's impractical or impossible and to outline the algorithms involved in each. For example, in sports, ladders are commonly used in playoffs where individual teams compete directly. But it's well-known that they can produce anomalous results because of early elimination (i.e., a good team gets eliminated in the first round and ends up at the bottom due to the misfortune of being paired against the top team while a not-as-good team rises because they played only weaker teams.) And, in search engines, where very large farms of machines are used in massively parallel searches, the strategy is to score any given matching page using heuristics that never consider how many other matching pages exist or even whether any exist at all. Msnicki (talk) 15:10, 4 July 2009 (UTC)