Author |
Message |
JUAN GARCIA OLMO (jgolmo)
New member Username: jgolmo
Post Number: 4 Registered: 6-2007
| Posted on Wednesday, November 26, 2008 - 11:02 am: | |
Dear Iyas, send me an email to my address ([email protected]) and I will send you a copy of the Math Discussion Appendix. Best regards |
Mariel elayne monrroy (mariel)
New member Username: mariel
Post Number: 3 Registered: 8-2008
| Posted on Friday, November 21, 2008 - 8:31 am: | |
Dear Howard Mark and JUAN GARCIA OLMO: Thank you for help. best regards. |
iyas (iyas)
Intermediate Member Username: iyas
Post Number: 20 Registered: 7-2007
| Posted on Friday, November 21, 2008 - 8:01 am: | |
dear JUAN GARCIA OLMO thanks for your answer if you please can you send a copy to me best regards |
JUAN GARCIA OLMO (jgolmo)
New member Username: jgolmo
Post Number: 3 Registered: 6-2007
| Posted on Friday, November 21, 2008 - 6:09 am: | |
Dear Iyas, you should contact with your local Perkin Elmer distributor. If you have problems, you can contact with me to send a copy of it. Best regards, |
iyas (iyas)
Intermediate Member Username: iyas
Post Number: 19 Registered: 7-2007
| Posted on Friday, November 21, 2008 - 5:57 am: | |
Dear all can you tell me when i can find the Math Discussion Appendix of the Quant+ Manual. to see the find information and equations about SEP and SEE ??? kind regards |
JUAN GARCIA OLMO (jgolmo)
New member Username: jgolmo
Post Number: 2 Registered: 6-2007
| Posted on Friday, November 21, 2008 - 4:55 am: | |
Dear Mariel, check the information included on the Math Discussion Appendix of the Quant+ Manual. You will find information and equations about SEP and SEE. Best regards, |
Howard Mark (hlmark)
Senior Member Username: hlmark
Post Number: 208 Registered: 9-2001
| Posted on Thursday, November 20, 2008 - 2:51 pm: | |
Mariel and Iyas - due to the history of how the mathematics for chemometrics developed, some equations came from different sources, such as Statisticians, Mathematicians, Chemists, Spectroscopists. They all knew the same equations, but knew them under different names, or created an equation and gave it their own name. This resulted in a situation where the same equation is called different names by different people, and the same name is used for different equations by different people. The important thing is the calculation that's done. After determining the calculation that you are working with, you have two choices: you can call it by a name that your target audience will recognize, or, preferably you can call it by the name that we are trying to standardize on. Standardized names are presented and defined by their corresponding equations in the Standard Practice E1655 of the ASTM, the American Society for Testing and Materials. We are tryng to persuade the chemometric and analytical chemistry communities, especially those using multivariate calibration methods, to utilize and promote the standard nomenclature for those calculations. By doing this, we hope to eventually get rid of the confusion attendant on the multiple names now being used by different people for the same quantity. That said, notice that in many cases, there is little difference in the practical application of some of these calculations. Playing a little loose with the terminology, all the terms mentioned, as well as some others, are calculated according to a similar formula. The formula has the form of the square root of a quotient. The numerator of the quotient, in the application, is invariably the sum of the squared differences between the instrumental values for the analyte, and the reference laboratory values. The denominator of the quotient is one of the following quantities: n-m n-m-1 (see more denominators below) where n is the number of samples and m is the number of Principal component factors or Partial Least Squares factors (as appropriate) used in a calibration. Other denominators in use are: n n-1 where: n is the number of samples used in a prediction set (also called a validation set) of samples. Note that the difference in the results of the formulas from what is nominally two different calculations, is equal to the quantity sqrt (n / (n-1)) or sqrt ((n-1) / n) Even if n (or n-m, perhaps)is as small as, say, 20, these formulas show that the difference between the different calculations is: sqrt (20/19) = sqrt (1.0526) = 1.026 or just over 2%. Given that nobody would ordinarily use such a small number of samples to do a chemometric calibration with, or a validation, we see that in practice, the effect of the differences between the different calculations is usually negligible in evaluating the performance of the calibration. It is more important, of course, to use the proper formuala for whether you are calculating calibration results or validation results. \o/ /_\ |
iyas (iyas)
Intermediate Member Username: iyas
Post Number: 18 Registered: 7-2007
| Posted on Tuesday, November 18, 2008 - 11:52 pm: | |
yes and i want to thank you for this question and i want to add is there an equation which give the relation OR a formula that give us the relation between SEP and rmsec and rmsecv AND SEE KIND REGARDS |
Mariel elayne monrroy (mariel)
New member Username: mariel
Post Number: 2 Registered: 8-2008
| Posted on Tuesday, November 18, 2008 - 10:07 pm: | |
Dear: Can you tell me which is the equation of the error of full cross validation for PLS that gives the software spectrum quant + (PERKIN ELMER). For example,In the spectrum quant, the calibration error is called SEE, but the equation that employed is the RMSEC. To full cross validation is called SEP.What is the equation? Thank you for help. |
|