Build multi-tenant RAG with Neon's database-per-user model — no nosy neighbors, max isolation, minimal costs

PostgreSQL WIDTH_BUCKET() Function

Summary: in this tutorial, you will learn how to use the PostgreSQL WIDTH_BUCKET() function for binning numeric data into discrete buckets.

Introduction to the PostgreSQL WIDTH_BUCKET() function

The WIDTH_BUCKET() function allows you to categorize numeric values into discrete buckets based on specified boundaries.

Here’s the basic syntax of the WIDTH_BUCKET() function:

WIDTH_BUCKET(numeric_value, lower_bound, upper_bound, num_buckets)

In this syntax:

  • numeric_value: The numeric value that you want to place into a bucket.
  • lower_bound: The lower bound of the range within which the numeric_value falls. It is inclusive.
  • upper_bound: The upper bound of the range within which the numeric_value falls. It is exclusive.
  • num_buckets: The total number of buckets to divide the range into.

The WIDTH_BUCKET() function returns an integer value that represents the bucket number into which the specified value falls.

If numeric_value is less than the lower bound, the WIDTH_BUCKET() function returns 0.

If the numeric_value is greater than or equal to the lower_bound but less than the upper bound of the first bucket, the WIDTH_BUCKET() function returns 1.

If the numeric value falls within the second bucket, the WIDTH_BUCKET() function returns 2.

This pattern continues until the numeric value falls into the last bucket, where the WIDTH_BUCKET() function returns the num_buckets

If the numeric_value is less than the lower_bound, the WIDTH_BUCKET() function returns 0. If the numeric_value is greater than the upper_bound, the WIDTH_BUCKET() function returns the num_buckets plus 1.

In practice, you often use the WIDTH_BUCKET() function for data binning tasks and histogram generation.

PostgreSQL WIDTH_BUCKET() function examples

Let’s explore some examples of using the WIDTH_BUCKET() function.

1) Basic WIDTH_BUCKET() function example

The following example uses the WIDTH_BUCKET() function to assign numeric values to three buckets (0, 10), (10, 20), and (20, 30):

SELECT
  WIDTH_BUCKET(-1, 0, 30, 3),
  WIDTH_BUCKET(0, 0, 30, 3),
  WIDTH_BUCKET(12, 0, 30, 3),
  WIDTH_BUCKET(25, 0, 30, 3),
  WIDTH_BUCKET(35, 0, 30, 3);

Output:

width_bucket | width_bucket | width_bucket | width_bucket | width_bucket
--------------+--------------+--------------+--------------+--------------
            0 |            1 |            2 |            3 |            4
(1 row)

In this example:

  • The WIDTH_BUCKET() function returns 0 for the value -1 because -1 is less than the lower bound (0).
  • The WIDTH_BUCKET() function returns 1 for the value 0 because 0 falls in the range of the first bucket (0,10).
  • The WIDTH_BUCKET() function returns 2 for the value 12 because 12 falls in the second bucket (10, 20).
  • The WIDTH_BUCKET() function returns 3 for the value 25 because 25 falls in the third bucket (20, 30).
  • The WIDTH_BUCKET() functions return 3 for the value 35 because 35 is greater than the upper bound (30).

2) Using the WIDTH_BUCKET() function with table data

We’ll use the film table from the sample database:

The following example uses the WIDTH_BUCKET() function to categorize the films into six buckets:

SELECT
  title,
  length,
  WIDTH_BUCKET(length, 40, 200, 6) AS bucket_number
FROM
  film
ORDER BY
  title;

Output:

title            | length | bucket_number
-----------------------------+--------+---------------
 Academy Dinosaur            |     86 |             2
 Ace Goldfinger              |     48 |             1
 Adaptation Holes            |     50 |             1
 Affair Prejudice            |    117 |             3
 African Egg                 |    130 |             4
...

3) Using WIDTH_BUCKET() function to generate histogram

The following example uses a common table expression (CTE) to generate the bucket numbers and then calculate the frequency of films falling into each bucket:

WITH buckets_cte AS (
    SELECT WIDTH_BUCKET(length, 40, 200, 6) AS bucket_number
    FROM film
)
SELECT
    bucket_number,
    COUNT(*) AS frequency
FROM
    buckets_cte
GROUP BY
    bucket_number
ORDER BY
    bucket_number;

Output:

bucket_number | frequency
---------------+-----------
             1 |       147
             2 |       203
             3 |       184
             4 |       194
             5 |       175
             6 |        97
(6 rows)

How it works.

CTE:

  • First, define a common table expression named buckets_cte.
  • Second, use the WIDTH_BUCKET() function to assign each film’s length into one of six buckets.

Main query:

  • First, select data from the buckets_cte.
  • Second, count the number of films for each bucket using the COUNT(*) function and group the result by the bucket_number using the GROUP BY clause.
  • Third, sort the result by the bucket_number.

4) Generate ASCII histograms

Based on the result set, you can generate a histogram in the application. But if you want to generate a histogram in psql, you can use the following query:

WITH buckets_cte AS (
    SELECT WIDTH_BUCKET(length, 40, 200, 6) AS bucket_number
    FROM film
),
histogram_cte AS (
    SELECT
        bucket_number,
        COUNT(*) AS frequency
    FROM
        buckets_cte
    GROUP BY
        bucket_number
    ORDER BY
        bucket_number
)
SELECT
    bucket_number,
    REPEAT('*', DIV(frequency,10)::int) AS histogram
FROM
    histogram_cte;

Output:

bucket_number |      histogram
---------------+----------------------
             1 | **************
             2 | ********************
             3 | ******************
             4 | *******************
             5 | *****************
             6 | *********
(6 rows)

How it works.

The query defines two common table expressions and generates the histogram using the asterisk character (*).

  • buckets_cte CTE – Assign films to buckets using the WIDTH_BUCKET() function based on their lengths.
  • histogram_cte CTE – Select data from the buckets_cte by grouping the frequency of the bucket numbers.
  • The main query – Select data from the histogram_cte CTE. We use the REPEAT() function to generate the histogram bars by repeating the asterisk (*) character a number of times which equals the integer division of frequency by 10. This is to scale down the histogram to prevent it from being too wide. To perform integer divisions, we use the DIV() function. Since the REPEAT function accepts an integer only, we cast the result of the DIV() to an integer using the cast operator (::).

Summary

  • Use the PostgreSQL WIDTH_BUCKET() function for binning numeric data into discrete buckets or generating histograms.

Last updated on

Was this page helpful?