LinkBack URL
About LinkBacks
Does anybody know how Excel decides what curve to put through the points on a scatter graph. I realise that one can fit a polynomial trendline but this is different to the curve that is drawn through the raw data.
Thanks
Does anybody know how Excel decides what curve to put through the points on a scatter graph. I realise that one can fit a polynomial trendline but this is different to the curve that is drawn through the raw data.
Thanks
Welcome to the board.
The curves are a Catmull-Rom spline.
Entia non sunt multiplicanda sine necessitate
Re: Curve fitting in scatter graphs
Thank you shg. Now I'll go and look up Catmull-Rome spline! If you have any suggestions for suitable websites I would be greatful.
Kind regards
Thanks again
You're welcome.
Shg, now I understand the principle behind the curve fitting but I have no idea how to produce the differential of the matrix equation. I have a graph of % released v time at 2 hour time points. I want too try to interpolate to 15minute time points. Can excel do this for me?
Regards
Why not just linearly interpolate?
Unless you can conceptualize why a spline is more appropriate to interpolation of the data because of some underlying phenomonology, it's just empty magnification (AKA manufacturing data).
Good question! The graph is clearly curved and with very wide time points. It might be easier if I attached a graph to show you? I'm not sure how to do that.
The issue that I have is that the 4th order polynomial does not look to fit the data as well as the Catmull spline. If I draw a straight line between the points it is both visually incorrect and does not fit the conceptual model
Of course you can post a workbook.
What conceptual model?If I draw a straight line between the points it is both visually incorrect and does not fit the conceptual model
I would like to post a workbook but could you tell me how to do that please?
The model is a diffusional model of drug from a tablet. The diffusion equations are well known but the complexities are changes of diffusional force (drug remaining), pH changes and changes in thickness of the membrane through which the drug is diffusing
The forum rules explain everything you need to know.
There is surely a pharmacological model that can be fitted that makes more sense than a random spline, of which there are many to choose from.
No simple model. The variables are unknown
Interesting problem. We are generally pretty good at programming things into Excel, but I don't know if any of us here are knowledgeable enough in diffusion models to help develop the model.
A quick internet search found several discussions and tutorials for different diffusion equations. This tutorial (http://www.rpgroup.caltech.edu/~nats...fusion_old.pdf ) seems interesting as one that is specifically targeting "readers from various backgrounds". If you are uncertain of the mathematics behind diffusion models, I would suggest that you spend some time with the differential equations.
At the end of your math work, I would expect you to end up with some kind of expression for "cumulative % released over time" (I find it interesting that this value goes above 100) as a function of the various input parameters you believe are important to the model. Once you have that expression, then you should be able to use Excel to perform a regression (linear or non-linear) on the data.
At this stage, if you are uncertain what the variables are, then I would suggest some time with the mathematics until you have some sense of what the variables should be.
Originally Posted by shg
Once again, my sincere thanks
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
Register To Reply
Bookmarks