curve-fit on segment

[blastguy]

Dave: Firstly, great software, beats my $2500 TecPlot. Would like to reinforce earlier requests: 1) option for 'curve fit on segment': most gage noise/drift problems my area only over part of data record. 2) log axis for 3D allows great way to present P-I diagram... which I probably dont need to explain to you after 30yrs in area....

[DPlotAdmin]

Thanks for your comments.

1) option for 'curve fit on segment': most gage noise/drift problems my area only over part of data record.

Have you only used Generate>Polynomial Curve Fit and not Generate>More Curve Fits? Or do I misunderstand what you're trying to do?

2) log axis for 3D allows great way to present P-I diagram...

On my list. Hopefully it won't be too much longer.

which I probably dont need to explain to you after 30yrs in area....

Actually you might - what's the third dimension? A probability of failure or... something else? I've never thought of presenting P-I's in 3D, definitely something to think about.

[blastguy]

1) my data records often have non-linear baseline drift starting at certain time into record, or burst of noise meaning there are different best-fits or averaging required for particular segments. Some parts are OK. I want to use prescribed fit or prescribed averaging for just selected segment of data, yet still work with whole record. Current curve-fit option seems only to allow for fitting of whole record (?). Therefore, I end up generating plot files for each segment of same record needing different treatment.
2) P-I diagram is really a 3D plot, where iso-damage lines are contours for response = height. Can use test or computational results for response at any P-I condition and use DPLOT to generate P-I diagram using contour option. However, P-I axes typically Log10.
I successfully used DPLOT for both above when others in project getting nowhere, however more pain than had to be.

[DPlotAdmin]

Using Generate>More Curve Fits (note: NOT Generate>Polynomial Curve Fit) there is a "Fit all points" checkbox. Uncheck that box. Is that not what you want?

[blastguy]

Yes, sorry, did find what I wanted in "other curve fits>etc" . However, I dont see the sense in having separate utility for the same options as in main curve-fit menu (eg, polynomial) if you just want to fit over a prescribed segment. Why not just have "from" and "to" in the main curve-fit options? have the full domain set as default. Trivial, but this is about the only thing I have found so far that has not been intuitive; I have not yet had to read the manual everything has been that easy/obvious.

[DPlotAdmin]

There is no sense to it now. At one time the "More Curve Fits" plugin wasn't available with the trial version and was meant to be an enticement to buy. But anybody can use it now, so the "Polynomial Curve Fit" command is for the most part redundant. It does have a couple of nice features that the plugin doesn't if your scales aren't linear, but those can fairly easily be taken into account with a few new equation forms in the plugin.