Orange Forum • View topic - Error using Distributions and 'no entry' sign above log reg

Error using Distributions and 'no entry' sign above log reg

A place to ask questions about methods in Orange and how they are used and other general support.

Error using Distributions and 'no entry' sign above log reg

Postby andyroberts1234 » Mon Jun 07, 2010 16:02

I am new to Orange and have been playing around. My overall aim is to create a nomogram for a logistic regression. I have managed to do this using the sample dataset. I then tried a made-up dataset "test.txt" saved as a tab-delimited file in the form below (but with 51 examples/cases)

a b c d
1 2 3 2
1 2 2 4
0 3 4 5
1 4 2 5

This loaded up ok using the file widget. Data table and Select Attributes seemed ok as well (1 discrete class variable, 4 continuous attribute variables). However when opening up the distribution widget , no graph displays and when I try to change the Variable I get the following error:

Unhandled exception of type AttributeError occured at 15:41:18:
File:, line 1406 in __call__
Code: self.f(**kwds)
File:, line 672 in setVariable
Code: self.graph.setVariable(self.attribute)
File:, line 135 in setVariable
Code: self.refreshVisibleOutcomes()
File:, line 287 in refreshVisibleOutcomes
Code: self.refreshPureVisibleOutcomes()
File:, line 237 in refreshPureVisibleOutcomes
Code: ckey.setRenderHints(QwtPlotItem.RenderAntialiased, self.useAntialiasing)
AttributeError: 'PolygonCurve' object has no attribute 'setRenderHints'

I also have a no entry sign above the icon for logistic regression (and other classifiers) when connected to the file. Is there a step I have missed that maybe was automatically there for the sample dataset?

Many thanks


Postby Ales » Tue Jun 08, 2010 9:26

The Distributions widget is fixed. You can update from SVN or wait for tomorows snapshot.

You can find out what about the logistic regression error by hovering the mouse over the shown icon above it on the canvas.

Return to Questions & Support