janezd
commented
3 years ago Thanks for this detailed and well-argued suggestion. And also for investigating the related proposals.
In the last link you posted, I suggested using Linear Projection. Fig. 8 from the paper looks like something you could plot with this widget. So -- would Linear Projection help? If not, could it be the starting point towards the widget you need?
ThirstyGeo
What's your use case?
In the world of geochemical data analysis, we often look at chemical element groupings spatially, because this helps geologists to understand inter-relationships of minerals (using their domain knowledge). One example of this is to look at chemical elements in 3D 'Euclidean' space where the axes are defined by three principal components - an example is shown HERE in Figure 8.
Further to this, geospatial workers often deal with data which has an X, Y, and Z coordinate. For example, sample data on the surface of the Earth, but also underground.
In both these two examples, geoscientists who use Orange struggle to examine their data in a 3D manner, and are forced to either (1) export data to a third party software which is not FOSS, or (2) write a 'good-enough' call to matplotlib to look at the data using the Orange widget.
What's your proposed solution?
A modified version of the Scatter Plot widget from Orange, or something like PLOTLY would probably be functional enough for typical geoscientist use, and student short courses.
Are there any alternative solutions? Back in 2011, this was investigated as part of the Google Summer of code: LINK; however, it was never implemented. In later feature request discussions (2018) the proponents didn't take time to argue the value of this idea: LINK