Scatter plot extremely slow on Windows with 1.12<scipy<1.15

[jsmariegaard]

At least two known installations of modelskill on Windows have extremely slow scatter plots (+1min) if show_density=True (which is the default). If show_density=False then speed is good (<1s) same if bins is set to a low number.

Python 3.13.0

matplotlib==3.9.2
mikecore==0.2.2
mikeio==2.1.2
modelskill==1.0.0
numpy==2.1.3
pandas==2.2.3
scipy==1.14.1
xarray==2024.10.0
ipykernel==6.29.5

[jsmariegaard]

I now tested different versions of scipy, and the problems start from version 1.13.0. The previous scipy version 1.12.0 (with NumPy 1.26.4) works fine.

[jsmariegaard]

The problem is this line:

modelskill/modelskill/plotting/_scatter.py

Line 659 in dd4122b

Z_grid = interpolate.griddata((xg, yg), hist, (x, y), method=method)

And the call to https://docs.scipy.org/doc/scipy/reference/generated/scipy.interpolate.griddata.html

--

The problem can be partly alleviated by using fewer number of bins:

cmp.plot.scatter(bins=40)   # default is 120

Or to not show the points with show_points=False or show fewer points show_points=1000.

[ecomodeller]

Is this problem specific to Windows?

[jsmariegaard]

Is this problem specific to Windows?

As far as I know. I don't have the same problem in WSL. I don't know why

[jsmariegaard]

Great news - seems to be resolved with the most recent version of scipy==1.15 released one week ago 😌