Add forecast functionality

[KKulma]

Pull the results of the short-term forecast available via Carbon Intensity API, see this endpoint.

[p-hunter]

Few notes on this one since I partially helped the first time, on things that could be useful for this issue:

• some sort of Weibull model for wind would be appropriate

• Hurdle/zeroinfl for solar

• Holt-Winters tends to work well for electricity demand with a ts object with very granular seasonality (I think the data are every hour... so 24 * 7, 24 * 30, 24*365 etc)

• when wind and solar are forecasted, the difference between demand - (wind+solar) should effect the carbon intensity

National grid ESO provide their own forecasts, but it's just a very large ANN with lots of layers, 64 nodes per layer. All theory, no practise, and little inference haha.

[KKulma]

Thanks for your comment, @p-hunter ! For this issue, I meant something much more trivial - pulling the results of the short-term forecast available via Carbon Intensity API, see this endpoint, for example. But of course, we could implement a more sophisticated algorithm in the future 👍

[p-hunter]

whoops! I must have read too deep into this then hahaha. But yes, one day . . .

[dr-harper]

This looks fun to have a stab at, let me see if I get time over the next month :)

Documentation here:
https://carbon-intensity.github.io/api-definitions/#get-intensity-from-fw48h

The approach from @p-hunter has been on my mind a bit, as the Carbon Intensity API only forecasts forwards 48 hours. Elexon provide 14 day wind forecasts although I haven't done any assessment of the accuracy of these. Either way, such work would probably be beyond the scope of this package and deserving of its own repository.