Data Smoothing

I'm sure it would be possible. Can you explain exactly how you want to smooth these data -- exactly what you mean by "merging" the peaks? Do you want some kind of moving average. Some kind of localized curve fitting? Can you show the results you expect in your sample file?

Once we understand exactly how you want to merge peaks, we can figure out the algorithm to accomplish that.

Localised curve fitting seems an appropriate method but I am not well versed in statistics. It is a little difficult to fully show what I fully mean. The gaussian smoothing here:

http://www.swharden.com/blog/images/smooth2.png

Shows what I need partially - where multiple peaks are represented as smoothed curves if they are adjacent/close to each other.

- Kayleigh

I haven't been able to fully wrap my head around Gaussian smoothing/filtering, but it also seems like you aren't fully decided that Gaussian filtering is what you want to use. At this point, I would suggest you decide what smoothing algorithm you want to use. Your sample plot shows Gaussian, Triangle, and Window. I'm also aware of moving average and Loess (local curve fitting using polynomials). There are probably others. I think most data smoothing algorithms will follow this basic skeleton:

1) Determine window size.
2) Obtain parameters for smoothing equation for the window specified in 1.
3) Using these parameters, calculate a smoothed value for y at the desire x.
4) repeat for the next desired window.

The big unkown is exactly what to put in step 2, which will depend mostly on what smoothing algorithm you decide you want. You might review this wikipedia article for a discussion of different strategies for data smoothing.
http://en.wikipedia.org/wiki/Smoothing

To illustrate the generic algorithm, here's a sample workbook using a "local polynomial" to smooth the data. I'm sure there are other ways to do it.

Controlling the desired "window" to be used for the smoothing function is done by changing the references in the linest() function.