I used in the past the version 8.1. It is now possible with the newer version of unscrambler to have more than 20 levels in categorial variabls?

Answer in version X, coming out end of the month this will be possible. 2 votes

I have a series of spectra from a protein complex. I want to know, how many species (monomers, complex) are in the samples for the following cases:1) For a series of spectra with increasing concentration of one of the components.2) For the case that I have a spectrum of a micture of the two components (with one component in excess) and both components separately. I want to know, how much of the components is bound and how much is still free.For both cases, the spectra are assumed to be linear combinations of the spectra of all species present in the sample.Can I do this using PCA/Unscrambler?Thanks a lot!Sabine

Answer Multicurve resolution (MCR) seems to be the best suitable method for this analysis as it gives both estimated signals and concentrations.In addition you have pure spectra that you can use as information background. Remember to suppress as much noise as possible as MCR is quite sensitive to noise.A PCA can also be used but the PCs may not be directly interpretable as spectral signal. 2 votes

Why is PLS-1 regression best MVA for spectral data? I've heard collinearity can be a problem in MVA. Spectra would have high collinearity - is it that PLS can 'deal with' collinearity better than other methods?

Answer You are right about collinearity. PLS deals better with it, as well as PCR: Principal Component regression. (This is compared to multilinear regression (MLR).)It is also a question of number of variables. In a spectrum you have many wavelengths (let's say 1000), if you want to create a MLR model you will need the number of wavelength plus 1 (1001) samples, which in many cases is very difficult to achieve. Finally the difference between PCR and PLS is that 1rst in PLS you use the variance most related to the Y variable to find the PCs (or latent variables) (in PCR you use the structure in X only), 2nd is that in PLS you can have several Y variables (PLS2). 1 votes

I've got few questions.1. what's the relationship among multivariate statistics, chemometrics and PCA2. i would like to know if i can use PCA to identify the sources of metal exposure in blood

Answer 1. Multivariate statistics encompasses methods such as PCA, PLS etc. Chemometrics is the application of multivariate statistics to chemical data.2. To answer this question, firstly, you need a method of analysis, such as spectroscopy, or a number of chemical/physical tests to be performed on representative samples. If the measures are related in some way to sources of metal, then PCA will show you and also provide information on the variables that are most related to the metal exposure. You can add categories to your data, based on known exposure levels to identify whether the method you are using can distinguish between the exposure levels. 1 votes

I have a formatting issue. If I cut and paste a graph from Unscrambler v9.2, how can I change the font size so that when I paste it into a word doc it is still legible?

Answer If you want to be able to edit font you can change the settings in Unscrambler. You access it from "File/System Setup" then in the "Viewer tab: Clipboard format:" select "Picture".Then in word or ppt you will be able to change things by right click and use the menu "Edit Picture". 2 votes

1 | 2 | 3 | 4 |