Match data series

[Solando]

Hi,

I want to use least squares method to compared some results with measured data. The only problem is that the points in the data series does not match. So my question is how can I do this?
I though maybe it would be possible to plot the lines and then use least squares on the lines themselves. However I have not been able to get this to work. Next I though about editing the data series to make the points match. As the resolution of the series are quite good, I don't think it will make "too much of a difference". So, does anyone know how to manipulate two data series to match the points on the x-axis by linear interpolation in Excel?

Very greatful for any help I can get

[Pauleyb]

I think this has had no real activity since there are so many questions...

Generally, least squares is used for a single data series. You get a linear equation that best fits the data. Are you planning on getting a linear equation for the first series and then another equation for the second series? Or, are you trying to get some linear equation using both data series (which I'm not sure how you would do that unless the data can easily be merged)?

If you just want to look at the two linear equations, use =LINEST().

But, you have the statement of "the points in the data series does not match." What does that mean?
One series of data is in seconds and the other is in minutes? If so, then just convert your data to a common measurement.
Or, did the sampling not occur at the same frequency or timing? If so, then it should not be an issue since your regression will give you an equation for all of the series which you can use to interpolate points where the sampling does not match.

Pauley

[Solando]

Hi Pauly,

I have a set of simulations that I would like to compare against a data serie with measured data. I would like to do this by determining which of the simulations that are closest to the measured by using least squares. The main issue in this context is, like you mentioned, that the sampling does not occur at the same frequency. As the lines are rather complex it will be difficult to find and equation for them, but I though maybe linear interpolation could be used to make the points in time (x-axis) match, and then use the least square method to find which of the simulations that lies closes to the measured data. In addition, the series are of different length, so my question is how to make the points in time match for the simulations and measured values.

Best regards,
Eirik

[Pauleyb]

Not knowing what your data looks like I can't really say the best method to interpolate between points. At the extremes, it could be perfectly linear or it could be completely random. I would recommend looking at your data and determining how well a linear interpolation fits your data. It may work, and it is easy to calculate.

As for comparing the two data sets, would CORREL be a better function?

Pauley