PCA and the related PLS is quite popular with statistians working with process data, and I have to admit I am a co-author on paper or two with the keyword PCA. However, it is oversold in terms of bringing insight to processes. Beware VIP plots!
Just a thought, not a sermon, from (probably) the only non-mathematical type who reads your stuff, but I got nothing out of this article. Most importantly, I do not get its practical political, ecomomic, or scientific importance in a world gone mad. Which is all I care about.
On several occasions in my environmental law career, I worked successfully in court and other contested settings with two statisticians. Our success was due to the fact that before they testified or submitted their final reports, I insisted they make abundantly clear to me and understandable to an intelligent layman a) their statistically-derived conclusions, b) how they got there, and c) the practical significance for the case. I made them understand that I had to understand all of that and it had to be made apparent to the final decisionmaker. So, our case preparation was a strenuous process of their explaining their statistical analysis to me and simplifying it for me, and my digesting, translating and presenting it in plain words, so the trier of fact got it.
Here, that is unncecessary. All that seems warranted is a brief statement of the bottom line: what the bad guys screw up, why they screw it up and what screwing up means for the public or for science, etc.
PCA and the related PLS is quite popular with statistians working with process data, and I have to admit I am a co-author on paper or two with the keyword PCA. However, it is oversold in terms of bringing insight to processes. Beware VIP plots!
TMI
Just a thought, not a sermon, from (probably) the only non-mathematical type who reads your stuff, but I got nothing out of this article. Most importantly, I do not get its practical political, ecomomic, or scientific importance in a world gone mad. Which is all I care about.
On several occasions in my environmental law career, I worked successfully in court and other contested settings with two statisticians. Our success was due to the fact that before they testified or submitted their final reports, I insisted they make abundantly clear to me and understandable to an intelligent layman a) their statistically-derived conclusions, b) how they got there, and c) the practical significance for the case. I made them understand that I had to understand all of that and it had to be made apparent to the final decisionmaker. So, our case preparation was a strenuous process of their explaining their statistical analysis to me and simplifying it for me, and my digesting, translating and presenting it in plain words, so the trier of fact got it.
Here, that is unncecessary. All that seems warranted is a brief statement of the bottom line: what the bad guys screw up, why they screw it up and what screwing up means for the public or for science, etc.
Sometimes you have to also write for your colleagues, as I was doing here.