Fix naming of GDP and population columns in SSP data aggregation within tools.costs

[measrainsey]

Small fix to total_gdp and total_population columns being switched in message_ix_models.tools.costs.gdp.process_raw_ssp_data()

In the message_ix_models.tools.costs.gdp.process_raw_ssp_data() function that pulls and combines SSP data in tools.costs, there is a block of code that converts the data in the Computer() into a pandas DataFrame, with each computer key being a column. However, the naming of the columns are switched (total_population and total_gdp). So, I just swapped them.

From what I can tell, this has zero impact on the actual costs calculations, as these DataFrames columns are not used anywhere (and are later even dropped). The calculation of GDP per capita happens before this DataFrame is created, here:

message-ix-models/message_ix_models/tools/costs/gdp.py

Lines 87 to 88 in 9bb0037

	# GDP per capita
	c.add(k_gdp["cap"], "div", k_gdp.base, k_pop)

And these values seem to be correct.

I picked up on this because I am calling using the output of message_ix_models.tools.costs.gdp.process_raw_ssp_data() function on its own elsewhere.

How to review

For @khaeru and/or @glatterf42 : Read the diff and note that the CI checks all pass.

PR checklist

• Continuous integration checks all ✅
• Update doc/whatsnew.

[khaeru]

I don't recall precisely, but I would guess that I put the total_population and total_gdp columns in place during work on #99 because you had other code further downstream that appeared to be using those data.

If that's since changed, and they're no longer needed, a better fix would be to remove them and the associated code entirely.

[measrainsey]

I don't recall precisely, but I would guess that I put the total_population and total_gdp columns in place during work on #99 because you had other code further downstream that appeared to be using those data.

If that's since changed, and they're no longer needed, a better fix would be to remove them and the associated code entirely.

Thanks Paul -- I am using the all of the columns in the output of message_ix_models.tools.costs.gdp.process_raw_ssp_data() (to make figures for writing/presenting), so for me it would be nice to keep the columns in there, even if they're not used for the module itself. But if it's causing too much clutter in the module code then we can also consider removing this part.

[khaeru]

I am using the all of the columns

Okay, then good to keep them. In that case I'd instead ask to expand the docs with half a sentence for these two columns, explaining precisely what is contained in each, so that others can maybe use them in the same way.

[khaeru]

The CI failures can likely fixed by cherry-picking 9e565c9 from #213.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 3 lines in your changes missing coverage. Please review.

Project coverage is 58.8%. Comparing base (cf687da) to head (33aa238).
Report is 6 commits behind head on main.

Files with missing lines	Patch %	Lines
message_ix_models/tools/wb.py	0.0%	3 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##            main    #219     +/-   ##
=======================================
- Coverage   58.8%   58.8%   -0.1%     
=======================================
  Files        194     194             
  Lines      15159   15161      +2     
=======================================
  Hits        8922    8922             
- Misses      6237    6239      +2     

Files with missing lines	Coverage Δ
message_ix_models/tools/costs/gdp.py	94.6% <ø> (ø)
message_ix_models/tools/wb.py	29.8% <0.0%> (-0.7%)	⬇️

[khaeru]

I am not sure why the .tools.wb cherry-picked changes are apparently not covered by the tests, since the corresponding tests do seem to run (although XFAIL).

However, LGTM and then continue with other work/clean-up, e.g. in #213 from which those changes were cherry-picked.