| Author | Message | ||
     Deepak Prem (Deepakp) |
Hi, I am a PhD student working on development of reliable calibrations for determination of biochemical variables, specifically, oil, moisture and protein content from Brassica whole seed samples. I am using a NIRS- 5000 equipped with NSAS version 3 spectral manipulations and regression development software. I have two questions for which I have not been able to find the answer: 1) Although I have done an extensive literature search, I have not been able to find any references for use of NIR (except for three references from the 10th international rapeseed congress, Canberra Australia, 1999) for determination of biochemical variables from Brassica seeds. Does anyone know of any such reports? 2) I have used PLS with cross validation for developing regression equations using 2nd derivative transformed spectral data. However, I have not been able to figure out the difference between Standard Error of Cross Validation (SECV) and Mean Square Error of Cross Validation (MSECV). Could someone explain the difference please? | ||
| David W. Hopkins (Dhopkins) |
Deepak, You did not mention the common name of the material you are working on. I assume you mean rapeseed when you say you are working on Brassica? Or what? Russ Tkachuk wrote a number of papers on rapeseed 1981 - 1988, measuring whole and ground samples as I recall. See JAOCS 65(3) 381-385 for his latest. There are many others in the CNIRS Bibliography, including papers by Salgo, Cowe, Dardenne, Daun and Biston to name a few of the other lead authors. Please contact me directly if you need more assistance. There has been a lot of work on rapeseed, and the biggest problem is the black seed coat of some varieties. The light varieties can even be done whole in transmission, I'm not sure that the dark ones can. I have Version 3.52 of NSAS, which I think is about the latest version of NSAS available. I think that they used MSECV to indicate what is presently referred to as SECV, and Std Error as presently called SEC. I hope this helps. Good luck in your studies! Best regards, Dave | ||
| Deepak Prem (Deepakp) |
Dear Dr Hopkins, I thank you for your prompt and valuable reply to my questions. I am working on rapeseed and mustard. Your comments and suggestions are indeed very helpful to me since I have been able to locate some of the papers of the authors mentioned by you. I must also thank you on clearing my concept on MSECV and SECV. You have mentioned in the reply that the big problem with rapeseed is the dark seed coat colour. Well, I gather this is true for transmittance mode NIR spectroscopy but I have been working in the reflectance mode and have had a fairly successful run so far. About a week ego I had written to you a brief synopsis of my work at the email address provided by this web site. I hope you have received the mail. Thanks again with warm regards deepak | ||
| David W. Hopkins (Dhopkins) |
Deepak, I am glad to hear that you have found more papers. As I recall, the earlier results were that reflection measurements could be used on the dark samples, even when transmission could not. No, I have not received the email from you. Please try again. Perhaps it got deleted with junk, please make sure it has a recognizable Subject. Best wishes, Dave |