A fuzzy join variable where we keep only the minimal value given by the by formula.
In theory we could automatically switch to "best join" if the result of the formula is numeric and not logical, but all of this is handled by fuzzy_join function so we can't intervene at this step.
We can do the cartesian product ourself but that's pretty much recoding fuzzy_join, which might not be a bad idea.
After this is done we still have the ambiguity of "best" join, best for grouping variables on the left/right ?
should the type of join imply it ?
or should it be an additional parameter ?
or should we only look at the table on the left ?
As it is a form of aggregation, it might fit better in eat, and then it is (I think) intuitive that the table on the left is the one we're grouping on to get these best values.
A fuzzy join variable where we keep only the minimal value given by the
by
formula.In theory we could automatically switch to "best join" if the result of the formula is numeric and not logical, but all of this is handled by
fuzzy_join
function so we can't intervene at this step.We can do the cartesian product ourself but that's pretty much recoding
fuzzy_join
, which might not be a bad idea.After this is done we still have the ambiguity of "best" join, best for grouping variables on the left/right ?
As it is a form of aggregation, it might fit better in
eat
, and then it is (I think) intuitive that the table on the left is the one we're grouping on to get these best values.