HISTOGRAM

Introduced or updated: v1.2.377

Generates a data distribution histogram using an "equal height" bucketing strategy.

Syntax

HISTOGRAM(<expr>)

-- The following two forms are equivalent:
HISTOGRAM(<max_num_buckets>)(<expr>)
HISTOGRAM(<expr> [, <max_num_buckets>])

Parameter	Description
expr	The data type of expr should be sortable.
max_num_buckets	Optional positive integer specifying the maximum number of buckets. Default is 128.

Return Type

Returns either an empty string or a JSON object with the following structure:

• buckets: List of buckets with detailed information:
  • lower: Lower bound of the bucket.
  • upper: Upper bound of the bucket.
  • count: Number of elements in the bucket.
  • pre_sum: Cumulative count of elements up to the current bucket.
  • ndv: Number of distinct values in the bucket.

Examples

This example shows how the HISTOGRAM function analyzes the distribution of c_int values in the histagg table, returning bucket boundaries, distinct value counts, element counts, and cumulative counts:

CREATE TABLE histagg (
  c_id INT,
  c_tinyint TINYINT,
  c_smallint SMALLINT,
  c_int INT
);

INSERT INTO histagg VALUES
  (1, 10, 20, 30),
  (1, 11, 21, 33),
  (1, 11, 12, 13),
  (2, 21, 22, 23),
  (2, 31, 32, 33),
  (2, 10, 20, 30);

SELECT HISTOGRAM(c_int) FROM histagg;

┌───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│                                                                                                              histogram(c_int)                                                                                                             │
├───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ [{"lower":"13","upper":"13","ndv":1,"count":1,"pre_sum":0},{"lower":"23","upper":"23","ndv":1,"count":1,"pre_sum":1},{"lower":"30","upper":"30","ndv":1,"count":2,"pre_sum":2},{"lower":"33","upper":"33","ndv":1,"count":2,"pre_sum":4}] │
└───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘

The result is returned as a JSON array:

[
  {
    "lower": "13",
    "upper": "13",
    "ndv": 1,
    "count": 1,
    "pre_sum": 0
  },
  {
    "lower": "23",
    "upper": "23",
    "ndv": 1,
    "count": 1,
    "pre_sum": 1
  },
  {
    "lower": "30",
    "upper": "30",
    "ndv": 1,
    "count": 2,
    "pre_sum": 2
  },
  {
    "lower": "33",
    "upper": "33",
    "ndv": 1,
    "count": 2,
    "pre_sum": 4
  }
]

This example shows how HISTOGRAM(2) groups c_int values into two buckets:

SELECT HISTOGRAM(2)(c_int) FROM histagg;
-- Or
SELECT HISTOGRAM(c_int, 2) FROM histagg;

┌───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│                                                  histogram(2)(c_int)                                                  │
├───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ [{"lower":"13","upper":"30","ndv":3,"count":4,"pre_sum":0},{"lower":"33","upper":"33","ndv":1,"count":2,"pre_sum":4}] │
└───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘

The result is returned as a JSON array:

[
  {
    "lower": "13",
    "upper": "30",
    "ndv": 3,
    "count": 4,
    "pre_sum": 0
  },
  {
    "lower": "33",
    "upper": "33",
    "ndv": 1,
    "count": 2,
    "pre_sum": 4
  }
]