Global Fits (parameter linked)

[Jbryant61]

Hi, I have attempted a simple equation to several sets of data. I am having trouble constraining one of the parameters to be linked for all data sets.

Is it possible to plot a variable number of data sets on a graph, depending on the number of columns in the data table?

Any assistance greatly appreciated.
Thanks
Jason

[RichardJ]

You have to fit all the data sets at the same time.

Richard

[Jbryant61]

superb..Thanks Richard.
Jason

[Jbryant61]

Richard, I think your global "y-axis" part is using values for the fit from the SSE results from my part. After deleting all the SSE part as you suggested, your graph no-longer works!

J

[Jbryant61]

Sorry, I know why now, as you said, I need to build the residual for all 4 y columns!!

[Jbryant61]

sorry one more thing. I am requested to express the Rsquared value. Instead of minising the residuals, could I equally well minimise the residual squared, and then use this for the denominator for the Rsquared. I have read that the numerator shoudl be the sum of the y values squared for my data, is this correct?

Thanks
Jason

[RichardJ]

On 2/24/2010 11:11:49 AM, Jbryant61 wrote:
>sorry one more thing. I am
>requested to express the
>Rsquared value. Instead of
>minising the residuals, could
>I equally well minimise the
>residual squared,

Not if you want it to be a least squares fit.

Minimize the residuals, and then just calculate the Rsquared separately.

Richard

[RichardJ]

On 2/24/2010 11:11:49 AM, Jbryant61 wrote:
>sorry one more thing. I am
>requested to express the
>Rsquared value.

Be aware that for Rsquared to have any statistical meaning the errors (i.e. the residuals) must be random and normally distributed. Yours clearly aren't, because the equations does not fit the data very well.

Richard

[Jbryant61]

Hi Richard, I had messed the "a" relation, it should all be related to the last column. The fits are a lot better. Im struggling with the R squared!

Jason

[RichardJ]

On 2/24/2010 11:47:13 AM, Jbryant61 wrote:
> Im
>struggling with the R squared!

It's a simple calculation

Richard

[Jbryant61]

Thanks.