defining line characteristics in fitted curves

[Sundaram Sridharan]

I would like to be able to set up logistic regression curve fit for a scatter plot and then select options: one line for each color and shape. And then I want to turn off display (of the fit line)  for one specific shape. Is that possible in Ironpython?

[David Boot-Olazabal]

Hi Sundaram,

You may find some snippets on our iron python library page: Iron Python library (search for line).

This one below, may be the starting point:

Kind regards,

David

[Sundaram Sridharan]

I am looking at the logistic regression class help page

https://docs.tibco.com/pub/doc_remote/spotfire/7.9.0/TIB_sfire-analyst_7.9.0_api/html/T_Spotfire_Dxp_Application_Visuals_FittingModels_LogisticRegressionFittingModel.htm

unfortunately this fitting model does not seem to have options to change line style, color etc let alone find a way to turn off a specific shape or color

[David Boot-Olazabal]

Hi Sundaram,

Can you share what you currently have, in terms of dxp file and iron python script?

Kind regards,

David

[Sundaram Sridharan]

In the file attached, I need to disable / make it invisible the fit line drawn through the cross shape in the second curve (in all curves).

logistic regression curve fit.dxp

Edited July 24 by Sundaram Sridharan

[David Boot-Olazabal]

Hi Sundaram,

I see. As you have multiple cross shapes in the second curve, and you have a line per shape, you'll get the two lines.
The upper curve only has one cross shape, hence no line is being drawn.

And disabling the line per shape, alters the complete line. I'm assuming that is not what you want either.
I'll ask my colleagues if there is a way around this.

Kind regards,

David

[David Boot-Olazabal]

Hi Sundaram,

We think there are a couple of workarounds possible.
The first one is not to use the Excluded data. That would give you the Included data and it's curve.

But if you want/need the Excluded data as well, you can color code it in white, so it doesn't show in the plot:

You can even get rid of the cross shapes, if you select the white cross shape in the 'Shape' section:

Kind regards,

David

[Sundaram Sridharan]

thanks - I thought about using white color - but I need to show the excluded points without showing the line through them. The only way I have been able to do so far is to have two different scatter plots, one like this and a duplicate copy where I draw a reference curve based on the calculated parameters for just the included points - but this slows down rendering/parameters export (to table) etc for very large data sets

Edited July 30 by Sundaram Sridharan