Feature request: Aggregation methods

[NunoSempere]

Something like

forecastNuno=0.9
forecastEli=0.6
forecastMisha=0.1
bestGuess=aggregate(forecastNuno, forecastEli, forecastMisha, method="geomOdds")

My guess is that having a few ones (geometric mean, geometric mean of odds, geometric mean of deviation from base rates), etc. would be relatively easy to program, and yet be appreciated by forecasters.

[NunoSempere]

Syntax would have to be figured out. Some options:

aggregate([f1, f2, f3, ...], method="geomOdds")
aggregate(f1, f2, f3, ..., method="geomOdds")
geomOdds([f1, f2, f3, ...])
geomOdds(f1, f2, f3, ...)

[OAGr]

Is "geometric odds" the same as "geometric mean"? If so, we could have a function geoMean (which we might want anyway). I assume it would accept an array. (I generally want to use easily typed function definitions, so arrays instead of unlimited paramaters)

That said, we might still want a function just for aggregation.

On syntax, I'd feel better about passing a hash, like,
aggregate([f1, f2, f3, ...], {method: "geomOdds"})

Related:
https://stackoverflow.com/questions/12826977/multiple-arguments-vs-options-object

[quinn-dougherty]

Seems like more of a standard library thing than a core language thing.

[NunoSempere]

Geometric odds is not the same thing, geometric mean of probabilities is (p1*...*pn)^(1/n), geometric mean of the odds would be the geometric mean of p1/(1-p1), p2/(1-p2), etc.

[NunoSempere]

Seems like more of a standard library thing than a core language thing.

I don't disagree