Add a source for convenient plotting of tabular data

[chpolste]

A basic source to simplify plotting of DataFrame(-like) objects:

import numpy as np
import pandas as pd
import earthkit.plots
import earthkit.plots.quickplot as qplt

Uses the x, y and z arguments to access columns, e.g.,

df = pd.DataFrame({"foo": [9, 16, 25], "bar": [3, 4, 5]})
qplt.line(df, x="bar", y="foo")

and falls back on the index for the x coordinate (if an index exists).

Also works with spatial data stored in a table, e.g.,

lat, lon = np.meshgrid(np.linspace(80, 20), np.linspace(-50, 50))
values = 10*np.sin(np.deg2rad(lon))*np.cos(np.deg2rad(9*lat))

df = pd.DataFrame({
    "lat": lat.flatten(),
    "lon": lon.flatten(),
    "values": values.flatten()
})

qplt.contourf(df, z="values")

or

subplot = earthkit.plots.Map()
subplot.contourf(df, z="values")
subplot.coastlines()

though it's not playing nice with quickmap as there is no datetime method on TabularSource.

I would like to automatically generate labels (for a line or the axes) based on the column names like pandas does with its built-in plotting routines but haven't been able to figure out how to do that yet.

[FussyDuck]

All committers have signed the CLA.

[JamesVarndell]

Hi @chpolste, this looks really great! I'm currently in the process of overhauling the earthkit-plots source objects, which should be released next week. Once that's out, we can tweak this tabular source and merge it in - should be a great addition.