Scenarios module

[znicholls]

Adds scenario module to OpenSCM. This gives us a bunch of scenarios with which we can test models and also makes it easy for first-time users to get a feel for how OpenSCM works without having to wrangle scenario data to get started.

Also includes a minor bugfix in ScmDataframe's interpolation method.

• Implementation finished
• Tests added
• Documentation added
• Example added (in the documentation, to an existing notebook, or in a new notebook)
• Description in CHANGELOG.rst added (single line such as: (`#XX <https://github.com/openclimatedata/openscm/pull/XX>`_) Added feature which does something)

[swillner]

Nice!
Just a few requests:

• Can you please clean the notebooks? ;) (we really need a check for that ;) )
• Can we have these CSVs as tables including only columns "Variable","Unit","Year", please?
• 20thcentury_emissions or historical_emissions (you also have some from the 18th century in there)?
• The historical emissions don't seem to be in the scenario module
• What about lazy-loading them?

[znicholls]

thanks @swillner! I think I've addressed all the comments now so ready for round 2

[znicholls]

we might want to re-write the timeseries notebook to make it more obvious that OpenSCM is super to easy to use, putting the details later in the notebook for those who are interested.

[swillner]

I would very much like the timeseries conversion notebook stick to the internal procedure and thus not use normally exposed interfaces... Probably we should have a thorough revision of all the notebooks once the package itself is in a releasable state.

So without those changes I think we are good to go. Maybe rename the column to "ParameterType" to be consistent with the (apparent) camel case names of the other columns (or use snake_case throughout)?

[znicholls]

Just to make sure I've understood, you want a notebook with only private interfaces exposed? Yes should be snake_case throughout, that's lazy by me.

…

On Thu, 23 May 2019 at 6:07 pm, Sven Willner ***@***.***> wrote: I would very much like the timeseries conversion notebook stick to the internal procedure and thus not use normally exposed interfaces... Probably we should have a thorough revision of all the notebooks once the package itself is in a releasable state. So without those changes I think we are good to go. Maybe rename the column to "ParameterType" to be consistent with the (apparent) camel case names of the other columns (or use snake_case throughout)? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#165>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFUH5G2BLZP6JMBD7ID7UQ3PWZGDDANCNFSM4HORCGAQ> .

[swillner]

Yes, it's about the internals, right? Anyway, let's leave the notebooks as they are right now and properly clean them once have a release candidate.

can you then also make the other columns snake case, please? ;)

[znicholls]

Yes, it's about the internals, right?

I thought it was about the fact that there are two different conventions with respect to timeseries and that they have to be handled differently.

Anyway, let's leave the notebooks as they are right now and properly clean them once have a release candidate.

Ok I'll just add a note in the top saying they're likely to be re-written once we have a release and put some suggestions about how re could re-write

[swillner]

Cool, just give me a ping and I'll merge ;)

[znicholls]

@swillner ping. A lot of the notebook changes are unavoidable cause of the case changes and the fact we can use scenarios now rather than downloading from iiasa in examples

[swillner]

Nice