d3.distinct() / d3.nest.distinct()

[mbostock]

It'd be nice to have a version of d3.nest for computing the number of distinct keys. Currently, you can do something like this:

var m = d3.nest().key(function(d) { return d.foo; }).entries(objects).length;

However, it would be faster to just compute the distinct count than to populate all the entries. That might look like this:

var m = d3.nest().key(function(d) { return d.foo; }).distinct(objects);

Alternatively, this could be provided as a d3.distinct(array[, f]) method. It's kind of nice that you can get the same functionality out of a nest operator, though. However, it's a bit weird since the nest operator allows multiple keys; do you want the number of distinct values for all keys? Or just the first key? Probably across all keys.

[mbostock]

Going to punt on this for now. I haven't decided whether we want a method that counts distinct values, or returns an array containing the distinct values. Likewise, there's a question of whether this should live on d3.nest, which is convenient if you already have a nest operator (and potentially minimizes surface area), but on the other hand, does not have obvious behavior if you want to return an array of distinct keys. Lastly, there's a question as to whether we want to determine distinctness by string coercion, or natural order comparison. And, whether the returned array of distinct values should be strings, or the original values before string coercion! Yikes!

[bddicken]

@mbostock I see this was closed awhile ago, but are you planning on adding something like this to d3? Or is there a function that does something equivalent? I for one would find it very useful. I ended up just implementing it myself in one of the vis projects I am working on, but I would think it a nice API feature.

[mbostock]

Well, there’s a slightly more concise version available now due to d3.map:

var m = d3.map(objects, function(d) { return d.foo; }).size();

[bddicken]

Ah yes, that works very well indeed! Thanks for the tip.