Add functionality to post-process prioritizr output

[sandra-neubert]

Add some functions that allow to process and visualise the output of a prioritizr run. This can include:

• Visualising prioritizr solution (also mentioned in #11 )
• Comparing different prioritizr outputs (maps that include no-regret areas, selected in one but not in other solution; plots comparing how well targets are met, some statistical measure (e.g. Cohen's Kappa) to show similarity)
• Visualising the differences between zones
• Tables of how well targets are met

Any other ideas @scabecks ?

[scabecks]

@diizhang do you have any ideas on this too? I can't recall what was done post-hoc in Colombia by Luizmar?

[diizhang]

Hi @scabecks, is this for the ELSA tool on UNBL? Please see my comments below: Visualising prioritizr solution

• Meaning the ELSA output maps? This is an existing function on the webtool.

Comparing different prioritizr outputs (maps that include no-regret areas, selected in one but not in other solution; plots comparing how well targets are met, some statistical measure (e.g. Cohen's Kappa) to show similarity).

• That will be great! For the plots on how well targets are met, can reference the radar chart on UNBC tool. But as we are using max coverage, the result is the % of features being represented in the final output maps instead of pre-set targets. There are budgets that we call area-based targets, that will be 100% meet in the result map.

Visualising the differences between zones

• Meaning the action zones in final ELSA map? That will be great if we can select and view separately. If it means the input zones, there is an existing function of visulizing zones restrictions in the input layers section.

Tables of how well targets are met

• same comments above for the % results. The existing tool has a result table like that, and downloadable as excel or csv.

[scabecks]

@diizhang these visualisation functions would be for creating nice outputs outside of the web-tools, for use in reports or other things (better than screenshots from the tools). They will not be modifications or additions to the shiny applications (that'd be another workload all together). Do we have examples of the post-hoc analyses Luizmar did to help guide us?

[diizhang]

@scabecks yes, here are the report of the post-hoc analyses: CR, Peru

[sandra-neubert]

Thanks @diizhang and @scabecks! Just double-checking most of the plots in the documents you sent, @diizhang, are screenshots and don't have R functions yet? Also, the max utility objective function does not necessarily need to represent all of the features. I know you use weights to increase the representation of your features and have for example Fig. 3 in the Peru document that shows how the zones contribute to the representation but even a 100% of the absolute representation across zones could in reality mean that a feature only has few planning units included in your solution, is that correct? For example, although it says 100% of High Integrity Forest is in the protection zone, how much of forest with high structural integrity is actually included in the solution (in PUs or area)? Sorry, maybe I'm missing something.

[scabecks]

@diizhang did Luismar share how he made those calculations and maps (most are clearly made in R with tmap)? Do you have his email to reach out, or could reach out yourself?

[diizhang]

Hi @scabecks @sandra-neubert , please find the code here: https://www.dropbox.com/scl/fo/1jtep24bad738vuv7mryz/ABRZlyGKUL0k9E_fnXywiQw?rlkey=5ebxo00vy7a5w85lamovx24p6&dl=0