Author |
Message |
Pierre Dardenne (dardenne)
Senior Member Username: dardenne
Post Number: 62 Registered: 3-2002
| Posted on Thursday, June 02, 2011 - 3:18 am: | |
Hello Nebojsa, I think the method proposed by Tony will lead to overoptimistic performances. It is more for a feasibility study than for the routine use of the models. The validations sets have to be independent of the calibration. The simplest way to set up the sets independently is to sort the spectra by date and make blocks of days, months or years. Pierre |
Richard Kramer (kramer)
Senior Member Username: kramer
Post Number: 27 Registered: 1-2001
| Posted on Friday, May 27, 2011 - 8:50 am: | |
One solution is to run your Foss spectrometer with the Symbion software (www.gosymbion.com). Using Symbion you are free to mix and match your choice of any chemometrics software which Symbion supports including Unscrambler. Symbion also provides nice capabilities beyond the standard Foss software. |
David Russell (russell)
Senior Member Username: russell
Post Number: 56 Registered: 2-2001
| Posted on Friday, May 27, 2011 - 8:09 am: | |
Back in the days when analyzers still ran DOS I used to convert Unscrambler(R) models to the NIRSystems .ca format for use with their DOS software. This required both formats to be documented and could be accomplished in a few days by a decent programmer. Unfortunately when the vendors moved to Windows(R) this required a significant development effort. So most vendors were unwilling to open their system to provide support for chemometric software developed by others without a business relationship. Additionally, the need for software to be "certified" or "validated" discouraged the use of third party features. So, as indicated by Peter's reply, the most practical approach is to duplicate your result as closely as possible in the chemometric package native to the analyzer. |
Peter Tillmann (tillmann)
Advanced Member Username: tillmann
Post Number: 23 Registered: 11-2001
| Posted on Friday, May 27, 2011 - 5:52 am: | |
Esther, welcome to this group. Theoretically it might be possible, but I doubt practically. Even if you put in lots of effort another limit is that e.g. the data pretreatments are not neseccarily identical in different packages, which is especially true for WinISI. WinISI supports only its own calibration files, which is also true for most / all (?) other chemometric packages. WinScan is the part in the old WinISI that controls the instrument. Similar to other vendors, which have chemometric packages and instrument control software integrated into a single software bundle, WinScan does not support foreign models files. (There might be exceptions to this rule of thumb.) Instrument control software from vendors, which have no chemometric packages themselves, frequently support model files from several chemometric packages. The solution is clearly to migrate the spectra to ISI and recalibrate with the knowledge you gained from Unscrambler. We do this frequently, but not to ISI, but from ISI to others. Peter |
Paladey, Esther (esther)
New member Username: esther
Post Number: 1 Registered: 5-2011
| Posted on Friday, May 27, 2011 - 3:29 am: | |
Hello all, I am new here, my question is: is it possible to convert calibration files from Unscrambler-software (9.1) (Camo) to winISI (2.0) software? My problem is that I use two software programs and in my opinion Unscrambler is more comfortable than our old winISI version. But our NIR runs only with this software. Thanks a lot in advance Esther |