Shameless self plug :)

https://cran.r-project.org/web/packages/eemR/index.html

Still not supporting PARAFAC but working on it.

I know that R is getting more and more popular. It is free and for Matlab you have to pay quite some money. However, for academics it is not really expensive and the packages you need to PARAFAC are free of charges available from the authors. So my choice was to go ahead with these and save some time for other work :-)

I really recommend the new toolbox from Murphy et al.

Murphy KR, Stedmon CA, Graeber D & Bro R (2013) Fluorescence spectroscopy and multi-way techniques. PARAFAC. Analytical Methods 5: 6557-6566

If you have negative values in your EEMs: Either the blank substraction (if already done) needs to be checked, or you may need to check the settings of the fluorometer. In case of very high or very low signals, the signal gets unstable and produces such artefacts (at least on Agilent/Varian machines, or on Perkin Elmer machines I know). If you have only very low negative values (

Gretchen, there are a few packages in R that deal with EEMs analysis, however PARAFAC has not yet been implemented in R. These packages mostly deal with preparing the data, peak picking (not an analysis I would advise), and fluorescence indices calculation.  As Klaus suggested, Matlab is the best option right now.  

Negative values could indicate a number of things and it is tough to give you a definite answer without seeing your data. 1) you may have a bad blank / bad DI water.  2) The lamp in your instrument may be going bad / the instrument is not properly calibrated and you are getting a lot of noise.  I have seen this in older instruments. 3) Small negative values are ok and DREEM can take care of those for you (I think). 4) Depending on the toolbox you are using make sure your constraints in all three modes are set to non-negativity.

Shameless self plug :)

https://cran.r-project.org/web/packages/eemR/index.html

Still not supporting PARAFAC but working on it.

Dear Philippe,

that would be a great job!

There are some new R packages. One is called 'multiway' which can do PARAFAC. You can see the description and tutorial of this package here: https://cran.r-project.org/web/packages/multiway/. Another is called "ThreeWay" https://cran.r-project.org/web/packages/ThreeWay/index.html , you can find reference here and there is also one document for this package: https://www.jstatsoft.org/article/view/v057i07/v57i07.pdf . And from the description, this package "PTAk" could also do PARAFAC: https://cran.r-project.org/web/packages/PTAk/.

But sorry, I don't use these R packages. I have used DOMFluor toolbox for about seven years and I'm used to it now (doing PARAFAC in Matlab combining DOMFluor and drEEM toolbox (some specific functions)).

Dear colleagues,

I am also working on bringing the PARAFAC from Matlab to R. The last days I was working hard on it and got some first results. I use the parafac function from the multiway package. I didn't do any corrections of the EEMs by myself. I got a Matlab workspace with the corrected EEMs. For more details please contact me.

Kind regards and a happy christmas to everybody,

Stefan

Has anyone attempted this in GNU Octave yet?

Dear Colloquies,

I am also interested in this topic, and would like to share my own experience on this topic: I was using Matlab for modeling tensor decomposition, which is indeed the most comfortable platform for general multi-way data analyses. Several tools and add-ins that come with MATLAB makes life even easier for processing multi-way data. One such tool is PLS toolbox from Eigenvector. However my personal experience with Eigenvector is not really good. I could say by experience that they are not friendly towards individual/ self-employed users. I learned this after e-mailing and calling them several times and literally begging that they validate my gmail for trial and purchase of their software. They even did not respond back! Registering with an institutional e-mail you might have a trial version of this package, later on subscribe for it. On the other hand current R-packages are premature and usually "case specific" for PARAFAC although there are promising libraries such as staRdom, multiway, PTAk, ThreeWay, etc. Besides, vendors of analytical instruments nowadays provide data acquisition and analysis tools which also include multiway and PARAFAC functionality.

HI Gretchen, I am working with EEMs and I am looking for a code to eliminate scattering and apply PARAFAC analysis. Could you develop the code? or can you help me with this? Gretchen Lescord

Use staRdom, includes drEEM and PARAFAC, incl scatter removal and interpolation. See answer and links above by Matthias Pucher.