Method to merge results plots

[camiloramirezgo]

Create a method to combine result plots as we do in latex for the publications. This method should be part of the OnStove class and allow one to take in different axes and position them.

[camiloramirezgo]

This is as far as I could get, as it is difficult to combine plotnine and matplotlib figures:

import matplotlib.pyplot as plt

fig = plt.figure(figsize=(12,9))

ax_dict = fig.subplot_mosaic(
    [
        ["bar", "map", "map"],
        ["hist", "map", "map"],
        ["ben", "map", "map"]
    ],
)

p = results.plot_split(cmap=cmap, labels=labels)
p.figure = fig
p.figure._themeable = {}
p._draw_using_figure(fig, [ax_dict['bar']])

p = results.plot_costs_benefits(labels=labels, height=1.5, width=2)
p.figure = fig
p.figure._themeable = {}
p._draw_using_figure(fig, [ax_dict['ben']])

p = results.plot_benefit_distribution(type='histogram', groupby='None', variable='wealth',
                                  cmap=cmap, labels=labels, height=1.5, width=3.5)
p.figure = fig
p.figure._themeable = {}
p._draw_using_figure(fig, [ax_dict['hist']])

results.plot('max_benefit_tech', ax=ax_dict['map'], cmap=cmap, legend_position=(0.85, 0.7),
             stats_fontsize=9, figsize=(7, 7),
             labels=labels, legend=True, legend_title='Maximum benefit technology', dpi=300,
             legend_prop={'title': {'size': 9, 'weight': 'bold'}, 'size': 9},
             rasterized=True, stats=True, stats_position=(0.85, 0.9), save_style=False)

Which outputs:

Something to notice is that the layout of the plotnine figures gets affected, this can be seen in the grids, the axis labels, and the legend of the figures getting messed up.

This requires more research, so we should leave this issue outside of the trieste milestone. If we really want this functionality in the future, maybe it is worth considering rewriting the plotnine figures using seaborn or pure matplotlib, this will allow for much easier merge of plots.

[babakkhavari]

Ok I agree, lets park this one currently