Skill plot by aux variable

[ecomodeller]

In an attempt to work with forecast skill, which seems possible to achieve using aux variables and no other changes, I wanted to plot the rmse as a function of lead time.
A bar chart works where each lead time is interpreted as a category, but a line plot is not correct.

[ecomodeller]

This is the problem, but I am not certain why this special case was there in the first place.

modelskill/modelskill/skill.py

Line 106 in eab3744

if not isinstance(df.index, pd.DatetimeIndex):

[jsmariegaard]

Cool

[jsmariegaard]

@jpalm3r - is this still a problem? (the line chart) - could you check?

[jpalm3r]

I think the problem is here in line 107. np.arange(nx) will always return an array that looks like [0, 1, 2, 3, ..., nx-1], which do not correspond to the lead time values when the time step is different than 1 (like in this case, which is 24).

I can fix it and make a PR.

[jsmariegaard]

Great - thanks

[jpalm3r]

Done in PR #446.

[jsmariegaard]

Closed by #446