Talk:Analytic hierarchy process

Move for deletion

I think this article goes beyond suck. It looks like someones just taken a "help" file for whatever this is and shoved it through machine translation. Sure it can be fixed, but is that :really: what we want?

From the Edit Box: Please do not create an article to promote yourself, a website, a product, or a business (see Wikipedia:What Wikipedia is not).

FrancisTyers 09:01, 29 Mar 2005 (UTC)

---

Hallo

my english is not very good, but I hope you understand me?

I tried translating my german topic to english language

Analytic Hierarchy Process

I apologise to You, if I have making mistakes in translation. Thank You very much for Your helping and deleting them quickly !? :-)

I am surprised, that in the English version of wikipedia not already its own, seperater contribution was written to the AHP? I have found nothing, why not, are there an explanation? The AHP originates nevertheless from North America? in addition I published a small google-statistics:

countries comparison in the employment of the Analytic Hierarchy Process (AHP)

by the way. I have creating the images myself !
need I doing something for copyright correctly publishing ??
please look the topic in my userpage of Rama

bytes are silver, talking togeter is gold:
--Demokrates 09:37, 29 Mar 2005 (UTC)

Ok, man, I didn't realise, just your formatting is all over the place, you need to fix up the capitalisation among other things... Perhaps it would be better if you don't know english that well to just start off with a small stub about what AHP is, rather than the whole thing...? FrancisTyers 09:45, 29 Mar 2005 (UTC)

I think you would be best off starting off with 'purpose' and 'definition of AHP', and leaving out the 'in practice' stuff, at least for now... FrancisTyers 09:46, 29 Mar 2005 (UTC)

Just a quick note here, I put the "delete" tag because I though that the article looked suspiciously like a advertisement for a commercial product. This of course does not mean to pre-judge of the "salvability" of the article. For what I have seen, though, there is a mathematical method which bears the name, so it would be nice to "re-deploy" the article toward a more mathematical approach... just a suggestion. Have fun ! Rama 09:53, 29 Mar 2005 (UTC)

---

Why Your destructive and radical reaction?!

Naturally deleting is the simplest, fastest and most comfortable way. however is there not always a better and more constructional alternative ?

If someone gives at lot trouble for making a little life to this page ? up to today's nothing was be reading here for the last 4 months?

Is this the best alternative: better nothing at all than something improving continually ? is this the international philosophy of wikipedia ?

Have You used or understand the method of AHP in Your own practice as yet ?

I think, the AHP is surely a very interesting method, but only someones do thinking, how explaining the method also for everyone simply understandably.

--Demokrates 10:18, 29 Mar 2005 (UTC)

Demokrates, sorry, I just meant to explain why I had previously put the tag on this page. Given the comments on this talk page, it seems that the article will indeed improve, given the apparent interest which arises around it. In this optic, the deletion tag is not really in order anymore -- in fact you are quite right to point it, I am removing it. Rama 10:31, 29 Mar 2005 (UTC)

Same here man! sorry, i wasn't trying to put you down, I was suggesting that instead of starting off with a huge page, it might be better to start off with a small page to allow people to contribute more easily. If you get some time, come on irc or something and talk to me, I'd be more than happy to discuss this with you, it seems that people easily get the wrong idea on these talk pages. FrancisTyers 19:45, 29 Mar 2005 (UTC)

---

Hi, it is not the nicest way to put down somebody who tried to fill the white spot about AHP in Wikipedia.He tried to make it easy for normal people not trained in mathematical approach for decision support!I do not want do read complicated treatises but I am grateful if somebody does make the attempt to define the AHP by means of a user-friendly application of this theory.And nobody would care giving a little help in rephrasing and fixing up the capitalisation for better understanding! --Felina 16:22, 29 Mar 2005 (UTC)

I agree that some kind of user friendly application might be nice, but first at least make a stub that explains it well in english. I did try and help with the capitalisation and grammar! Look at the edit history! I'm not saying it should be deleted, just improved, I don't have the necessary understanding to fix it properly. FrancisTyers 19:45, 29 Mar 2005 (UTC)

Start again

Rama said he translated the German article,[1] but if he did, it's improved rather a lot since then. Wikipedia articles are not supposed to be how-tos (which are useful external links of course). This article needs starting again from scratch. Rd232 10:37, 4 October 2005 (UTC)

I agree. I suggest the article should describe the problem (multi-criterion decision making) then the process of forming a hierarchy and assigning pairwise preferences in the range 1 to 9 (or, rather, 1/9 to 9) for how well each alternative satisfies a given criterion, and for the relative importance of the criteria. It should then show how AHP combines the pairwise comparisons into an overall preference (for example, see http://citeseer.ist.psu.edu/santillo00early.html or http://msdn.microsoft.com/msdnmag/issues/05/06/TestRun/default.aspx) Accommodating a group of decision makers should also be described. (That's in Saaty's 1980 book, but I can't find such a description on the web at the moment.) I would like to see a worked example - but showing the math. The current "software demo" might follow. --Jrvz 16:26, 11 January 2006 (UTC)

comparison to other methods and shortcomings

There are several key shortcomings of AHP and it is quickly being replaced by other methods. I'll add some material on that and perhaps that will address some of the issues others brought up in here. However, the article still needs further expansion and explains almost nothing about how AHP works. I added a bit more to the explaination and removed the confusing material which really explained nothing about it.

The article is still a How-To, and still uses German illustrations. Rd232 ^talk 10:18, 26 February 2006 (UTC)

Hey, by inserting a factual, non-commercial introduction, maybe I've "started it over." Lou Sander 01:51, 16 May 2006 (UTC)

I changed some of the claims back to something accurate. AHP cannot replace Navier Stokes equations for optimal wing designs nor can it replace standard probability theory and equations in statistics an Actuary uses to compute risks. Saying that AHP can be used to evaluate these things is only something a person who only knows AHP and understands little about other mathematical evaluation methods. For example, if you want to compute the trajectory of the Space Shuttle and the International Space Station so that they meet in orbit, you can't use AHP. There are laws of physics that govern this. Thats the problem with AHP proponents. They dont realize that many of the problem they use AHP for already have proven mathematical solutions.

Wikipedia, above all else, cannot introduce factual errors.Hubbardaie 14:46, 11 June 2007 (UTC)

Ad Hominem

Someone keeps stating that challenges are made by "people who don't understand (AHP)". This is a petty ad hominem. The researchers who have found these fundamental errors understand the problems much more than the typical user of AHP. This person is also removing citations that show specific mathematical proofs of the claims. If you cannot read and understand the proof for yourself, refrain from claiming that those who prove your position wrong simply don't understand it. This person also continues to insist that AHP would add value to formal mathematical optimization methods in various fields where, by definition, thier solution is already optimal. It does not "miss the point" that AHP cannot change an optimal answer and, therefore, cannot possibly add value. This is a person whose only exposure to any decision analytic methods at all is AHP and therefore, has no understanding of why it doesn't work on a variety of problems. Hubbardaie 13:32, 29 June 2007 (UTC)

RozannSaaty

Rozann Saaty is the wife of Thomas Saaty, who is the inventor of AHP. She writes with an obvious COI and POV. I suggest to her that she not delete citations to known mathematical proofs that are critical of AHP. I suggest leaving in those proofs and responding to them in a substantive manner.Hubbardaie 02:16, 3 July 2007 (UTC)

It now appears that TSaaty himself (Thomsas Saaty, I presume) as well as his wife is intent on making this article an advertisement or vandalizing what others write. Mr. and Mrs. Saaty are invited to discuss changes on this page. It appears they do not even want anyone to know that there are widely recognized theoretical problems with AHP. Instead of trying to hide it, I recommend they include rebutals to these issues to contribute to a neutral point of view on the page.Hubbardaie 11:36, 3 July 2007 (UTC)

I have spoken with the Saatys and have suggested to them to check the Discussion page.

Not such a good article

This article started out as a pretty poor one, then it got a lot better. Then the "criticisms" section got out of hand, and it's on its way down again. For example at the moment, all three of the article's references apply to the criticisms section. The article is NOT about "what's wrong with AHP," but about AHP itself. There's nothing wrong with briefly mentioning criticisms, but it's VERY wrong when the criticisms outweigh the article itself. I'm not an AHP subject matter expert, but we sure could use one to straighten out this article. Lou Sander 23:13, 30 July 2007 (UTC)

I disagree entirely that it got worse with the criticisms section (I wrote most of the criticisms). In fact, it was a huge improvement. Without the criticisms it was just a advert for a method that has been severely discredited among researchers in the decision sciences. If your only criticisms of the criticisms section is that it takes up too much of the article and all of the references are from the criticisms section, then I would say the criticisms section is the is the only complete part. Write more of the intro section and more of an explaination of AHP. There is plenty to say. You could write an entire section just on the Eigenvector math used for the consistency coefficient. Knock yourself out. But it would be a big backward step to reduce the size of the criticisms section just because not enough was written about the rest.Hubbardaie 00:19, 31 July 2007 (UTC)

I'm not surprised that you think that your own work in criticizing the subject of the article is appropriate and well done. I don't agree on either point, however. The article is about AHP. Lou Sander 01:58, 31 July 2007 (UTC)

Of course it is well done. Any objective judge (which you apparently are not) would agree. If you would bother to check, you would find that just about every article on any decision-making method, theory, or heuristic has a criticisms section. Since AHP has some specific well-known criticisms, it would be incomplete without it. Where else would you put the criticisms about the AHP method? Under cajun cooking recipes? And, as you already admitted, the criticisms section is the only part of the article that has references. Sounds like the problem is not about the criticisms section, but the rest of the article. I agree the rest of the article is not nearly as good as what I wrote. But you are free to improve it. How about taking a crack at the Eigenvalue consistency coefficient explaination? Hubbardaie 03:43, 31 July 2007 (UTC)

Actually, upon further reading, I see that the article (contrary to your claim above) does have other references. Just not as many as the criticisms. Again, feel free to expand the rest of the article to catch up with the quality and completeness of the criticisms section.Hubbardaie 03:47, 31 July 2007 (UTC)

Sorry, but IMHO the criticisms section isn't even a coherent piece of writing, let alone something appropriate for an encyclopedia (look at the first sentence, for example). Maybe too many cooks have spoiled the jambalaya. Lou Sander 04:05, 31 July 2007 (UTC)

The only problems I have with it now is that the criticisms section seems to be a series of compromises between pro-AHP people and my desire to include citations of its limitations (as do most articles about such methods). As the citations make clear, these are widely known shortcomings and any article on AHP would be incomplete without them. Sort of like having an article on homeopathic medicine without any mention that it has been challenged by the established medical community. I just modified the tone of the first sentence. The remaining main criticisms are that 1) AHP is often used as a solution where there are already proven optimization models (where it would, of course, give a less than optimal answer) 2) rank reversal flaws and 3) indifferent criterion flaws. What's so incoherent about that? Feel free to change it. Past edit wars on this article resulted only when people were trying to entirely remove any reference to the theoretical shortcomings of AHP. As long as you don't remove those critical points, you should feel free to change whatever you like. Then I can evaluate the coherence of your writing.Hubbardaie 04:22, 31 July 2007 (UTC)

IMHO the article is definitely getting better, but still has a long way to go. The main part of the article is still pretty light. Also the first and second paragraphs of the "criticisms" section mostly consist of weasel words. The last paragraph isn't about AHP, but about alternatives to it. (Many have objected to the spiciness of cajun cuisine. French, Indian, and McDonald's are much less spicy and much more widely enjoyed.) Lou Sander 12:38, 31 July 2007 (UTC)

I think the relevance of the alternatives here is that often users of AHP are assuming they need to use AHP because it is a "soft" problem with no mathematical solution. Freguently, this is not at all the case and they simply were not aware that there is a solution. One of the most common errors in application is applying it to uncertain and risky decisions where the risk would normally be quantified like an actuary would quantify it. Being unaware of this, the AHP user assumes that AHP must be the only way to answer the question and, instead, produces a suboptimal answer. You can see remants of this discussion in the paragraph about use of AHP for wing design or insurance risk. A previous pro-AHP user insisted that AHP gives the correct answers on these very specific formalized problems (of course, it could not). If the widely held assumption behind the use of a cajun recipe was that there is no other way to eat fish and rice, it would be valid for an encyclopedic article to point out the alternatives. Even more appropriately, if the assumption behind the use of homeopathic medicine was that there was currently no cure for a disease when in fact there was, then the mention of the real cure would be relevant in that article.

Good stuff, though it's hard (for me) to imagine people making important decisions that have specific solutions (ex: wing design) who are not aware of the specific solutions. Maybe "criticisms" could have something like "One criticism of AHP is that sometimes it's used where other, better, methods are available," etc. etc. etc. "There have also been criticisms of the underlying mathematics: etc. etc. etc." Lou Sander 16:10, 31 July 2007 (UTC)

I use that example because I think it makes the problem a bit more clear and obvious. But many people are using AHP on a variety of portfolio prioritization and valuation problems that, just like wing design, already have optimal solutions. In those cases, the users really are unaware that methods like Modern Portfolio Theory and various methods from operations research already have optimal solutions for what they are working on. These tend to be users in management who are not necessarilly aware of more advanced methods or believe that somehow their problem is less "quantifiable". Often, this assumption is in error. I think it orgiinally read something closer to what you suggested but after several back-and-forth edits with pro-AHP people, it evovled into what it is. When I get a chance I'll try to clean it up as you suggest but it may be a while (I just had my new book come out this week and I'm doing a lot of promotion-related events). If you have some specific approaches in mind for explaining this, take a shot at it.Hubbardaie 17:43, 31 July 2007 (UTC)

Can you give us some references to people wanting to use AHP for inappropriate purposes like wing design? It seems as though that would be a problem with those people, and not with AHP. If we could see some details, we could better evaluate the criticisms that are based on this. The same is true with the portfolio stuff. (Who are any of us to decide how people prioritize their portfolios?) Lou Sander 17:59, 23 August 2007 (UTC)

I second that idea. The criticisms section has a lot of gratuitous potshots and material that does not apply to the Analytical Hierarchy Process. Good Cop 18:34, 23 August 2007 (UTC)

The wing design example came from a previous edit war on this article. It previously did claim it could be used exactly for that. More importantly, AHP is routinely used for portfolio prioritization which already has a formally proven optimization processes for many types of portfolios. I also replaced the reference to Egon Brunswik. He's dead. It's not a commercial.Hubbardaie 14:08, 25 August 2007 (UTC)

1) Archived edit wars are not suitable sources for material in the encyclopedia. Please remove this material or provide a citation from a reliable source.

I didn't say it was. I was explaining how it ended up there after a series of edits between two sides. Feel free to remove it.Hubbardaie 21:30, 25 August 2007 (UTC)

Done. Editors who post uncited or improperly cited material are often given the chance to remove it themselves. But our patience doesn't endure forever. Good Cop 23:44, 25 August 2007 (UTC)

2) Egon Brunswik may or may not be dead, but the assertion that he produced a method superior to the subject of this article is unsupported. Please remove this material or provide some citations from reliable sources.

The assertion that Brunswik was dead was not made to argue that the method was superior. It was made to argue against the claim that the entire paragraph was a "commercial" (which would be hard to argue given that this is, now, a historical figure that has no relationship with existing commercial enterprizes). I added a citation regarding the relative accuracy of the two methods.Hubbardaie 21:30, 25 August 2007 (UTC)

3) The "commercial" is the unsourced material, inserted by Hubbard, about Hubbard's own Applied Information Economics. You can look at that article and its discussion page to see various complaints about conflict of interest, spamming, inappropriate posting, etc. involving Hubbard.

First, it appeared you labeled the entire paragraph a commercial, not just the AIE reference. Second, I encourage anyone to research the complaints in detail and the subsequent refutations of the complaints (I think you will find that every one has been addressed to the satisfaction of the original source of the complaint). Furthermore, As I've stated, I specifically use the HubbardAIE screenname to disclose my identity. What is inappropriate is always a matter of debate in wikipedia and I've also made it clear that I concede if someone insists something is inappropriate. But I wonder if "Goodcop" might be so forthcoming with his identity (for all I know, he is simply a competitor of mine). Hubbardaie 21:30, 25 August 2007 (UTC)

I should also point out that the Applied Information Economics method is published in public domain (a point I also made in the discussion page of that article). In that sense, it is no different than Saaty contributing to the AHP article. In fact, I think it is very likely that at least some of the past editors of this article that were insisting their were no flaws in AHP worked for one of firms peddling AHP software (of course, they don't reveal that as I do). I think there are degrees of being a "commercial" where a vendor promoting a specific trademarked/patented/copyrighted product (which I did not do) would be the most flagrant type. People who published work in an area and who offer their own work as sources (while I concede the point that it could still be considered a type of self-promotion) should not be lumped in with the first type.Hubbardaie 21:58, 25 August 2007 (UTC)

4) The "commercial" in the AHP article is also inaccurate/misleading, in that it claims that AHP "considers some things to be immeasurable." As far as I can see, AHP makes no claims about anything being immeasurable. It offers assistance where matters are regarded as such by its users. Because of the conflict of interest, lack of citations, and misleading nature of the "commercial," I am going to remove it. 71.245.188.100 17:57, 25 August 2007 (UTC) (The previous material is from me. I forgot to sign in. Good Cop 18:03, 25 August 2007 (UTC))

I suppose in the same sense that astrology itself could be said not to make paranormal claims but its users have certain paranormal beliefs, you might be correct, but it would be a stretch. It may be true that AHP does not specifically state what is immeasurable. The users simply presume the problems AHP is applied to have had no other formally derived optimal answer. In the spirit of comprimise, I'll concede that point. Hubbardaie 21:30, 25 August 2007 (UTC)

Perez, et al.

The criticisms section has a claim that these fellows have found a "strong theoretical problem" with AHP. What they actually claim is a "potentially strong theoretical problem," which is quite a different thing. Also, the citation leads to a paper that says nothing about its source. It seems to be an unpublished student paper of some kind. The analysis seems to be original research. IMHO such a paper is not a reliable source for criticism of something as solidly established as AHP. Coupled with the removal of "potentially," and the presence of the apparent original research, this is probably fatal. Unless these things are quickly fixed, the Perez stuff should be removed. Good Cop 00:06, 26 August 2007 (UTC)

I don't see the two issues as somehow cumulative in the decision to remove the reference. They are separate. The first issue is dealt with by simply putting "potential" in front of "strong" in the first sentence. The second, separate, issue of whether it is original research is entirely up to the interpretation of what constitutes a reliable source. When I search this one it appears to be part of a university doctoral program in Spain and I see many references to it on the web. It is refered to as a "work document" in the program at this source:

http://translate.google.com/translate?hl=en&sl=es&u=http://www.uv.es/bibsoc/GM/data/alcalcddt.html&sa=X&oi=translate&resnum=1&ct=result&prev=/search%3Fq%3D%2522Another%2BPotential%2BStrong%2BShortcoming%2Bof%2BAHP%2522%26start%3D10%26hl%3Den%26sa%3DN

Clearly, any article that makes uncited assertions could be guilty of violating NOR, but that is not the issue, here. The research is not original in wikipedia but it is citing a source that is not yet in a peer reviewed journal. But this university apparently posts the working papers of its doctoral candidates on the web in many locations. Does this suffice as publication? If you insist that the NOR rules clearly specify what qualifies as published and that papers of doctoral students posted by their university on the web is not among them, then it should be removed. But I bet there are many citations of websites and non-peer reviewed "vanity" journals throughout wikipedia. If anything other than citations of peer-reviewed, first-teer journals or books published by one of the major publishers counts as original research, then wikipedia has a lot of clean up ahead of it.Hubbardaie 01:18, 26 August 2007 (UTC)

You need to fix your distortion of the claims made in the paper, and you need to show readers where the paper came from. Also, you need to delete your personal embellishments/interpretations/distortions of what the paper actually says. Otherwise, people will just remove it. Good Cop 02:20, 26 August 2007 (UTC)