Enable using a column’s aggregator in Python expressions

[daniel-g-gagnon]

Column aggregators should be usable in Python expression.

Take for instance this data I was working with

province	population
Québec	9110616
Ontario	16176977
Colombie-Britannique	5719961
Alberta	4980659
Manitoba	1507330
Saskatchewan	1253569
Nouvelle-Écosse	1080418
Nouveau-Brunswick	859839
Terre-Neuve-Labrador	545464
Ile-du-prince-Édouard	180029

I would like to be able to add a sum on population, then be able to add a new column with the expression =population / population_sum to get the ratio from each province. Right now I have to hard code the sum in my expression.

Or maybe have an aggregator object, where I could get the value at aggregators.sum.population?

[midichef]

That's a great suggestion. I've had the need for that many times myself, without quite realizing there should be a simple way to access the aggregator in an expression.

It's not difficult to implement. The only part I don't know how to handle is to cache the value to avoid recomputing the aggregator once on every row.

Meawhile, a workaround is that you can store the aggregator value using z+ (memo-aggregate). That stores your sum in vd.memory with the key population_sum. You can access the stored value in expressions as vd.memory.population_sum or vd.memory['population_sum'].

So if you put in your .visidatarc: mem=vd.memory
you could access the value with mem.population_sum.

[saulpw]

We could put the aggregated value in vd.memory automatically. And possibly add vd.memory to the lookup chain. The only difficulty is that vd.memory is global, whereas these aggregated values are per-sheet. Maybe we could have a per-sheet memory?