Elasticsearch histogram, multiple interval types?

Question

Using Elasticsearch histogram functionality, i can put various 'ranges' of data into a bucket by specifying an interval. In this case '50':

Price: 0-50 50-100 100-150 150-200 200-250 etc

This works fine, but this returns an awfully long list of buckets. What I'd prefer is:

0-50 50-100 100-200 200-400 400-1000 1000+

Or something of the likes. Is it possible to tell ES what intervals (/ranges) it should return?

Answer 1

You need to use the numeric range aggregation, which allows you to specify exactly which intervals you want, such as this:

{
    "aggs" : {
        "price_ranges" : {
            "range" : {
                "field" : "price",
                "ranges" : [
                    { "to" : 50 },
                    { "from" : 50, "to" : 100 },
                    { "from" : 100, "to": 200 },
                    { "from" : 200, "to": 400 },
                    { "from" : 400, "to": 1000 },
                    { "from" : 1000 }
                ]
            }
        }
    }
}

This should give you exactly what you expect.

Answer 2

Documentation explicitly says about histogram aggregation that

It dynamically builds fixed size (a.k.a. interval) buckets over the values.

What I can think of is that in order to reduce number of buckets you may apply logarithmic scale (or any other non-linear scale, e.g. square root, that will give enough granularity for your particular dataset) to values using script option:

{
    "aggs": {
        "prices": {
            "histogram": {
                "field": "price",
                "script": "Math.log10(_value)",
                "interval": 1
            }
        }
    }
}

This will give buckets with keys 1, 2, 3,… that stand for original values within intervals [0;10), [10;100), [100;1000),…

By applying reverse function (10^x in this case) to the keys on client side you can restore original scale.