Code to reproduce JOSS figure

[FLomb]

JOSS's reviewers asked for the inclusion of the code to reproduce the illustrative Figure we have in the paper (see: openjournals/joss-reviews#6418 (comment)). This PR addresses this.

However, the PR is currently based on the old plotting functionality with some customisation. As discussed during RAMP's development meetings, it'd be better to produce a similar Figure with the newest plotting functionality. I tried doing so via the shadow plot functionality, but I struggled to customise it to the degree needed. For example, I did not manage to pass kwargs related to the ax object or to the colours of the lines. Perhaps @mohammadamint may know how to do it. If not possible, we may as well stick to this custom code.

[FLomb]

I'm not sure why this file ended up in the commit; I shouldn't have changed anything there and for some reason git diff does not return any difference between my commit and the joss-paper branch for this file.

[Bachibouzouk]

Apparently the image generated must be slightly different, this is the only line highlighted (once you commited git diff will not show the diff anymore), you can do git diff HEAD~1 to get the diff of the last commit.

[Bachibouzouk]

You can use

uc = UseCase("JOSS-paper-figure", users=[LMI, LI])
uc.initialize(num_days=1)
plot = uc.generate_daily_load_profiles(
    cases=[1,2,3,4,5]
)
plt.rcParams.update({'font.size': 22})

fig, ax1 = plot.shadow( figsize=(16, 8))
ax1.legend(['5-day average', 'Single days(s)'])
ax1.margins(0)

ax1.set_xticks(ax1.get_xticks()[1:])
import matplotlib.dates as mdates
myFmt = mdates.DateFormatter('%H:%M')
ax1.xaxis.set_major_formatter(myFmt)

ax1.set_ylabel('Normalised power (kW/kW-Peak)', fontsize='large', labelpad=20)
ax1.set_xlabel('Time (hours:minutes)', fontsize='large', labelpad=20)

To generate a figure similar using the new plotting capabilities

[mohammadamint]

You can use

uc = UseCase("JOSS-paper-figure", users=[LMI, LI])
uc.initialize(num_days=1)
plot = uc.generate_daily_load_profiles(
    cases=[1,2,3,4,5]
)
plt.rcParams.update({'font.size': 22})

fig, ax1 = plot.shadow( figsize=(16, 8))
ax1.legend(['5-day average', 'Single days(s)'])
ax1.margins(0)

ax1.set_xticks(ax1.get_xticks()[1:])
import matplotlib.dates as mdates
myFmt = mdates.DateFormatter('%H:%M')
ax1.xaxis.set_major_formatter(myFmt)

ax1.set_ylabel('Normalised power (kW/kW-Peak)', fontsize='large', labelpad=20)
ax1.set_xlabel('Time (hours:minutes)', fontsize='large', labelpad=20)

To generate a figure similar using the new plotting capabilities

@Bachibouzouk Thanks :-). Sorry for my delay in replying!

[FLomb]

Ok, thanks for your guidance. I've updated the figure accordingly. What we get is an almost identical result, except for colours. I reckon there's no way to change the colours of the shadow function because they are hardcoded, aren't they? Should I commit a change to the function that allows for colour customisation, too?

[Bachibouzouk]

function because they are hardcoded, aren't they?

You can get the plotted lines from the axis handle
https://matplotlib.org/3.1.0/api/_as_gen/matplotlib.axes.Axes.get_lines.html

[FLomb]

function because they are hardcoded, aren't they?

You can get the plotted lines from the axis handle https://matplotlib.org/3.1.0/api/_as_gen/matplotlib.axes.Axes.get_lines.html

Thanks for the tip! I've updated the code with custom colours. Now, the figure looks as expected; I should update the paper with it because it's still marginally different.

[Bachibouzouk]

Thanks @FLomb this looks neat

I added a seed to the UseCase so the figure is reproducable