| Author |
 Topic  |
|
|
TobiasKe
Germany
2 Posts |
 Posted - 10/02/2018 : 06:36:50 AM
|
Version: OriginPro 2017 (64-bit) SR1
Operating System: Windows 10
Hello,
I have worked a while with Origin now, but got stuck now on the following Problem:
I have a 3-D Scatterplot, with the data being on a grid x1,x2,x3; y1,y2,y3 with the corresponding z values z11, z12, z13 ... z33.
So far no problem and fitting worked also perfectly fine. What I'd like to do now, and can't find how to do that, is to basically "fit" the projections.
In the 3-D Scatterplot I'd like to fit a function to z11,z12,z13 (x=x1 fixed, y = y1 .. y3) and I'd obviously like the fit to show up in the scatterplot (as a projection would be also ok, I guess). Kind of a "line-plot" or "non-linear fit" (but not a surface approximation) to a subset of my data.
What I tried so far: Fitting an normal function in the 3D plot, which did not work out, except that it messes up my Z-axis labels. Fitting a Surface plot with the x not used in the explicit form, thus making the fit idenepndent of x. I hoped to project the fit later, but the fitting did not work, since all my data is collinear (duh... I know, that's the point...). Adding a function graph to the Scatter plot is also not possible (greyed out), otherwise I could have done the fits seperately and then just added them to the graph.
To make fitting (probably) easier, I also added subsets of the data to the graph (so only z11, z12, z13 with the corresponding x and y values). So I have the data I'd like to fit separated from the rest of the data. But no improvements...
An now I'm out of ideas. Having 3 graphs is of course possible, but not the desire solution.
Can someone tell me if, or ideally, how I can solve this problem?
Best regards
TobiasKe
|
|
|
Hideo Fujii
USA
1582 Posts |
Posted - 10/02/2018 : 1:09:30 PM
|
Hi TobiasKe,
I wonder if the following Waterfall plot with fitting lines can be waht you are looking for, or not.
I have made this waterfall by making subgroups at each Z(depth) to have both source scatter plot and its fitted curve.
You can take a look this blog about subgrouping:
http://blog.originlab.com/graphing/how-to-make-a-waterfall-graph-with-subgrouped-plots
Do you think this type of graph can serve you?
--Hideo Fujii
OriginLab |
 |
|
|
TobiasKe
Germany
2 Posts |
Posted - 10/05/2018 : 04:01:27 AM
|
Hi Hideo Fujii,
Thank you for this answer. The plot kind of helped, but did not fulfill all my needs. Especially since I also wanted to do a fit z(y,x=xi) and a fit z(x,y) (I forgot to mention, sorry); Also fitting 'inside' the Graph didn't seem to work (fit ok, but it wouldn't show up, probably wrong Z-Position). And horizontal droplines aren't available in a Waterfall plot.
BUT:
Dragging an existing fit into the Graph did work (after adding the Z-Position).
An so, thanks to you, I think I found a solution that works for me. It doesn't sound intended, but it does the job....:
1. Do all the fitting in seperate plots that are only 2 dimensional ( z(x,y=yi), z(y,x=xi) )
2. Do a Waterfall plot of my original data as you described it
3. Remove all Data from the Graph
4. Readd the Data in XYZ Fromat by "Drag&Drop" as a scatterplot
5. Add all the Fits in XYZ Format by "Drag&Drop" as a scatterplot
6. Add Combined Fit in XYZ Format by "Drag&Drop" as a scatterplot and change it to a surface plot
7. Layout stuff (Drop lines, Projections, everything works)
(Important to consider: the z-axis is horizontal now, so some columns might need renaming)
Interestingly, I can add Data as a scatterplot via Drag&Drop to a Waterfall plot. But adding any Data to a Scatterplot (after creating a Graph via selecting my Data -> Draw -> 3D Scatterplot) doesn't work for me, at least not with Drag&Drop. (Is this inteded?)
I admit, it's not a nice solution. But it's a workaround.
Thank you for your helpful input again.
Best Regards
TobiasKe
|
Edited by - TobiasKe on 10/05/2018 04:03:08 AM |
|
|
|
Hideo Fujii
USA
1582 Posts |
Posted - 10/05/2018 : 11:21:19 AM
|
Hi TobiasKe,
Your work-around sounds great. Is it possible to paste a screenshot of your sample graph here, so that
I or someone here (developer?) may be able to come to implement a right tool, or to write a blog or tutorial
with a solid procedure?
Again, thank you very much for bring your issue to this forum.
--Hideo Fujii
OriginLab |
|
|
| |
Topic |
|
|
|
Nonlinear Fit to a Subset of a 3-D Scatterplot