Comments on: Toeplitz Matrices and the Stahle Tree Ring Network http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/ by Steve McIntyre Mon, 20 May 2013 05:49:54 +0000 hourly 1 http://wordpress.com/ By: Kriging on a Geoid « Climate Audit http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-240289 Thu, 26 Aug 2010 18:24:25 +0000 http://www.climateaudit.org/?p=2898#comment-240289 [...] is far more knowledgeable, and has posted about the topic frequently on CA. See, for starters, “Toeplitz Matrices and the Stahle Treering Network”, 3/22/08, “Antarctic Spatial Autocorrelation #1″, 2/20/09, “Steig Eigenvectors and [...]

]]> By: harold http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-141739 Wed, 26 Mar 2008 23:53:48 +0000 http://www.climateaudit.org/?p=2898#comment-141739 Repost, different wording. I think there is a typo in the post.The author of HORSESHOE FOR MULTIDIMENSIONAL SCALING is de Leeuw (not de Leuuw as is mentioned 3 times).

]]> By: Patrick Henry http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-141738 Mon, 24 Mar 2008 16:25:17 +0000 http://www.climateaudit.org/?p=2898#comment-141738 “If the facts don’t fit the theory, change the facts.”
–Albert Einstein

]]> By: Falafulu Fisi http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-141737 Mon, 24 Mar 2008 14:57:42 +0000 http://www.climateaudit.org/?p=2898#comment-141737 Kernel PCA is available in the freely available Matlab SPRT (Statistical Pattern Recognition Toolbox).

Also, there are also more references and links to free software (R, Matlab, C, Java …) related to kernel methods found in the following link. The link contains the software for the book itself but the site lists other softwares as well.

- Kernel Methods for Pattern Analysis

]]> By: Falafulu Fisi http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-141736 Mon, 24 Mar 2008 14:46:17 +0000 http://www.climateaudit.org/?p=2898#comment-141736 Steve, have you tried any non-linear PCA? There are some free Matlab codes for kernel PCA available from the internet.

]]> By: harold http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-141735 Sun, 23 Mar 2008 22:55:20 +0000 http://www.climateaudit.org/?p=2898#comment-141735 Very interesting post.
Btw the author of A HORSESHOE FOR MULTIDIMENSIONAL SCALING is Jan de Leeuw (not Leuuw).

]]> By: Robert Wood http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-141734 Sun, 23 Mar 2008 22:16:21 +0000 http://www.climateaudit.org/?p=2898#comment-141734 Although my mind numbed a bit reading this post (it’s been a long time since college) it appears that there was a lot of eye-balling and intuiting in data selection, and the Taminoesque attempts at justification are justy baffle-gab – blinding with science.

You’re a patient man Steve.

BTW Apparently, sea temperatures and atmospheric temperatures are not increasing.

]]> By: Steve McIntyre http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-141733 Sun, 23 Mar 2008 13:01:07 +0000 http://www.climateaudit.org/?p=2898#comment-141733 #17. My conclusion is that the spatial correlation in the Stahle network is quite strong and suggests that one can use this to advantage in checking for outliers i.e. are there subsets that do not cohere to the spatial correlation pattern e.g. perhaps the Graybill sites.

If the NOAMER network is also held accountable to having spatial autocorrelation, the improbability of the Graybill chronologies increases even more. What are the odds that the top 15 HS-shaped sites (in terms of 20th century difference of mean) drawn from a network with high spatial autocorrelation should all be drawn by the same guy? Does Graybill have a magic flute or a flawed instrument?

If these were satellite measurements and one class of instrument had a major drift relative to other instruments, the instrument class would show up in a Mannian PC analysis – but that wouldn’t show that the instrument was valid. I’m going to do a post using this illustration.

]]> By: Craig Loehle http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-141732 Sun, 23 Mar 2008 12:52:44 +0000 http://www.climateaudit.org/?p=2898#comment-141732 This is why, if there is any hope of using proxies, it must be done by reconstructing local temperatures and then combining them. Averaging z-scores (tree rings or Thompson’s ice cores) or doing a global/regional PCA or something simply does not make sense.

]]> By: fred http://climateaudit.org/2008/03/22/toeplitz-matrices-and-the-stahle-tree-ring-network/#comment-141731 Sun, 23 Mar 2008 11:54:27 +0000 http://www.climateaudit.org/?p=2898#comment-141731 Could anyone give a simple explanation in a few lines of what this means? Is it something like that the readings are more correlated with geographical position than anything else…?

]]>