Forward looking: Separate activities that provide a budget from activities that declare they don’t provide a budget

[andylolz]

Recently (in #531) there was a change to include activities with the @budget-not-provided attribute in the forward looking “current activities with budgets” figure.

The wording here seems a bit odd, since these activities are in fact without budgets (in fact, it’s explicitly declared that they do not provide a budget). I wonder if it might be clearer to show two figures (one in brackets) to differentiate:

1. activities that do provide a budget, from
2. activities that either provide a budget, or explicitly declare that they do not (via the @budget-not-provided attribute)

[andylolz]

NB I’m not necessarily looking for immediate action on this! Just raising it here for posterity.

[Ocre42]

This was a change made to sort #454, as stated: "If budget-not-provided is present, the Dashboard logic should be that all required budgets exist."
I did include a flag to notify that if it is raised, some or all activities include the budget-not-provided attribute.

[andylolz]

The flag idea is great. I think a bracketed figure (as suggested above) would be even better.

In general, the dashboard uses flags to indicate a possible issue with the data (e.g. transaction dates in the future). It uses a bracketed figure to differentiate results for two different methodologies, which is the case here (i.e. excluding vs including budget-not-provided). Also, the two things being flagged here (budgets at multiple hierarchies and budget-not-provided) are orthogonal, so sometimes you may have to flag both things. That’s a bit confusing, and not how flags work elsewhere.

So I think a bracketed figure would be more idiomatic, and also more informative. Fortunately, your IATI-Stats code provides all the data you need to do this.

Separately, I think the code in #531 is a bit wrong anyway – but this isn’t apparent yet because there are no tests, and budget-not-provided is not widely used.

[andylolz]

Separately, I think the code in #531 is a bit wrong anyway

Specifically: if budget-not-provided is present for any activity, the code flips to use forwardlooking_activities_with_budget_not_provided values. This will probably result in the wrong answer. It should instead be adding the forwardlooking_activities_with_budgets and forwardlooking_activities_with_budget_not_provided numbers together.

UPDATE: this comment has been addressed in #555 and #556.

[andylolz]

From #555 (comment), I gather the decision here is to mark this issue as a wontfix.

The new methodology is apparently: If a publisher doesn’t provide a budget but gives a reason, that’s equivalent to providing a budget.

From the data user’s perspective, these things are not equivalent. If the goal of the publishing statistics is to measure against the needs of the data user (for whom forward-looking data is known to be beneficial), then I think this methodology change should be reconsidered.