Plot scalar results

[jnnr]

Introduces a new rule "plot_scalar_results" that draw scalar result bar plots for single scenarios. You can choose to stack the bars by setting stacked=True in plot_grouped_bar. Needs rl-institut/oemoflex#45

To test it, make sure to use the mentioned branch on oemoflex.

• Generate some scalar plots with snakemake -j1 results/toy-scenario/plotted/scalars.
• To test the integration in the report, run snakemake -j1 results/toy-scenario/report.

TODO

• Nicer appearance
• Plot other variables (see comment below)
• Check if and how it can be combined with the plot functions of joined scenarios. Done - there is only one script for both.
• Problem: Include in reports fails if file is not present. How can we deal with that? Will be addressed in New plots are missing in report #66

Address separately

• all variable costs by region and technology
• all storage losses + output by carrier and region (This would need a new functionality: Plotting different var_names together)

[jnnr]

Short brainstorm on what plot we may need:

• all flows into electricity bus/heat bus etc
• all capacities/invest of power into electricity bus/heat bus etc

[MaGering]

I get a MissingInputException in line 343 with the following error message running
snakemake -j1 results/toy-scenario/plotted/scalars:

Missing input files for rule plot_scalar_results:
results/toy-scenario/postprocessed/scalars.csv

Also, when I run snakemake -j1 results/toy-scenario/postprocessed, it leads me to AttributeError: 'NoneType' object has no attribute 'is_potentially_variable:

ERROR: Constructing component 'Transformer.relation_build' from data=None
    failed: AttributeError: 'NoneType' object has no attribute
    'is_potentially_variable'
Traceback (most recent call last):
  File "scripts/optimize.py", line 50, in <module>
    m = Model(es)
...

[jnnr]

I get a MissingInputException in line 343 with the following error message running snakemake -j1 results/toy-scenario/plotted/scalars:

Missing input files for rule plot_scalar_results:
results/toy-scenario/postprocessed/scalars.csv

Resolved in ec05d1e. Now snakemake should trigger a model run when you try to plot results that have not been created.

Also, when I run snakemake -j1 results/toy-scenario/postprocessed, it leads me to AttributeError: 'NoneType' object has no attribute 'is_potentially_variable:

ERROR: Constructing component 'Transformer.relation_build' from data=None
    failed: AttributeError: 'NoneType' object has no attribute
    'is_potentially_variable'
Traceback (most recent call last):
  File "scripts/optimize.py", line 50, in <module>
    m = Model(es)
...

Could not reproduce this. Looks like a component of the model is missing some data?

[MaGering]

Running snakemake -j1 results/toy-scenario/report leads me to an error because pandoc is missing as package. It should be added to install_requires in setup.py right?

[jnnr]

Running snakemake -j1 results/toy-scenario/report leads me to an error because pandoc is missing as package. It should be added to install_requires in setup.py right?

You do not install pandoc via pip, so it should not be listed in setup.py. Need to find another way to define these requirements.

[jnnr]

After installing pandoc I get the following error message running snakemake -j1 results/toy-scenario/report: report_exit_status_99.txt

Yes, the report fails if some images are missing. This is an open todo (see first post).

[SabineHaas]

You could install pandoc from github https://github.com/jgm/pandoc using the master branch or a specific version (tag).
For a "how-to" see oemof-B3 setup.py where oemoflex ist installed from github, as well.

[jnnr]

I am not sure if that works? Other than oemoflex, pandoc is written in haskell and needs to be compiled. (https://github.com/jgm/pandoc/blob/master/INSTALL.md#getting-the-pandoc-source-code)

[SabineHaas]

@jnnr you have asked my feedback concerning the plots. Thank you, first of all, for improving the plots! They already look very nice.
I have tested the plots with scenario base-2050, toy-scenario ran into an error.

To my mind, my comments are not so urgent that they have to be faced in this PR or for the first release.

scalars:

• I like that for the y-axis a convenient unit seems to be chosen automatically! A bit strange looks the '0 Whor0 W` to my mind. Maybe there's a solution to force 0 to adapt the unit of the next higher value.
• It would be useful for readers to add values to the stacked plots. I have added two examples below from Dena-Leitstudie 2021 and Prognos 2021.
• For the model I think B/BB and the heading are quite right, as they are related to input data and output data. However, for reports I think it would be nice to actually write the full region name and remove the heading as this will be separate in a report. My suggestion is to add a second set of plots for reports without heading and with full region names.
• For your information: I only got plots for the electricity bus. I guess plots for the other busses will be added in the future.

dispatch:

• snakemake -j1 results/base-2050/plotted/dispatch gives me ValueError: need at least one array to concatenate in oemoflex/tools/plots.py, line 344, y = np.vstack(y)

---

---

Address separately

* all variable costs by region and technology

* all storage losses + output by carrier and region (This would need a new functionality: Plotting different var_names together)

* Check if and how it can be combined with the plot functions of joined scenarios

Should this be addressed in another PR? Then you could move these points to a new issue.

[jnnr]

@SabineHaas: Thanks for your review. I opened a new issue #154 to collect all ideas that are still open.