postgresql return 0 if returned value is null

Question

I have a query that returns avg(price)

  select avg(price)
  from(
      select *, cume_dist() OVER (ORDER BY price desc) from web_price_scan
      where listing         


        

Answer 1

I can think of 2 ways to achieve this:

• IFNULL():

  The IFNULL() function returns a specified value if the expression is NULL.If the expression is NOT NULL, this function returns the expression.

Syntax:

IFNULL(expression, alt_value)

Example of IFNULL() with your query:

SELECT AVG( price )
FROM(
      SELECT *, cume_dist() OVER ( ORDER BY price DESC ) FROM web_price_scan
      WHERE listing_Type = 'AARM'
        AND u_kbalikepartnumbers_id = 1000307
        AND ( EXTRACT( DAY FROM ( NOW() - dateEnded ) ) ) * 24 < 48
        AND IFNULL( price, 0 ) > ( SELECT AVG( IFNULL( price, 0 ) )* 0.50
                                     FROM ( SELECT *, cume_dist() OVER ( ORDER BY price DESC )
                                           FROM web_price_scan
                                           WHERE listing_Type='AARM'
                                             AND u_kbalikepartnumbers_id = 1000307
                                             AND ( EXTRACT( DAY FROM ( NOW() - dateEnded ) ) ) * 24 < 48
                                         ) g
                                    WHERE cume_dist < 0.50
                                  )
        AND IFNULL( price, 0 ) < ( SELECT AVG( IFNULL( price, 0 ) ) *2
                                     FROM( SELECT *, cume_dist() OVER ( ORDER BY price desc )
                                           FROM web_price_scan
                                           WHERE listing_Type='AARM'
                                             AND u_kbalikepartnumbers_id = 1000307
                                             AND ( EXTRACT( DAY FROM ( NOW() - dateEnded ) ) ) * 24 < 48
                                         ) d
                                     WHERE cume_dist < 0.50)
     )s
HAVING COUNT(*) > 5

• COALESCE()

  The COALESCE() function returns the first non-null value in a list.

Syntax:

COALESCE(val1, val2, ...., val_n)

Example of COALESCE() with your query:

SELECT AVG( price )
FROM(
      SELECT *, cume_dist() OVER ( ORDER BY price DESC ) FROM web_price_scan
      WHERE listing_Type = 'AARM'
        AND u_kbalikepartnumbers_id = 1000307
        AND ( EXTRACT( DAY FROM ( NOW() - dateEnded ) ) ) * 24 < 48
        AND COALESCE( price, 0 ) > ( SELECT AVG( COALESCE( price, 0 ) )* 0.50
                                     FROM ( SELECT *, cume_dist() OVER ( ORDER BY price DESC )
                                           FROM web_price_scan
                                           WHERE listing_Type='AARM'
                                             AND u_kbalikepartnumbers_id = 1000307
                                             AND ( EXTRACT( DAY FROM ( NOW() - dateEnded ) ) ) * 24 < 48
                                         ) g
                                    WHERE cume_dist < 0.50
                                  )
        AND COALESCE( price, 0 ) < ( SELECT AVG( COALESCE( price, 0 ) ) *2
                                     FROM( SELECT *, cume_dist() OVER ( ORDER BY price desc )
                                           FROM web_price_scan
                                           WHERE listing_Type='AARM'
                                             AND u_kbalikepartnumbers_id = 1000307
                                             AND ( EXTRACT( DAY FROM ( NOW() - dateEnded ) ) ) * 24 < 48
                                         ) d
                                     WHERE cume_dist < 0.50)
     )s
HAVING COUNT(*) > 5

Answer 2

use coalesce

COALESCE(value [, ...])

The COALESCE function returns the first of its arguments that is not null.  
Null is returned only if all arguments are null. It is often
used to substitute a default value for null values when data is
retrieved for display.

Edit

Here's an example of COALESCE with your query:

SELECT AVG( price )
FROM(
      SELECT *, cume_dist() OVER ( ORDER BY price DESC ) FROM web_price_scan
      WHERE listing_Type = 'AARM'
        AND u_kbalikepartnumbers_id = 1000307
        AND ( EXTRACT( DAY FROM ( NOW() - dateEnded ) ) ) * 24 < 48
        AND COALESCE( price, 0 ) > ( SELECT AVG( COALESCE( price, 0 ) )* 0.50
                                     FROM ( SELECT *, cume_dist() OVER ( ORDER BY price DESC )
                                           FROM web_price_scan
                                           WHERE listing_Type='AARM'
                                             AND u_kbalikepartnumbers_id = 1000307
                                             AND ( EXTRACT( DAY FROM ( NOW() - dateEnded ) ) ) * 24 < 48
                                         ) g
                                    WHERE cume_dist < 0.50
                                  )
        AND COALESCE( price, 0 ) < ( SELECT AVG( COALESCE( price, 0 ) ) *2
                                     FROM( SELECT *, cume_dist() OVER ( ORDER BY price desc )
                                           FROM web_price_scan
                                           WHERE listing_Type='AARM'
                                             AND u_kbalikepartnumbers_id = 1000307
                                             AND ( EXTRACT( DAY FROM ( NOW() - dateEnded ) ) ) * 24 < 48
                                         ) d
                                     WHERE cume_dist < 0.50)
     )s
HAVING COUNT(*) > 5

IMHO COALESCE should not be use with AVG because it modifies the value. NULL means unknown and nothing else. It's not like using it in SUM. In this example, if we replace AVG by SUM, the result is not distorted. Adding 0 to a sum doesn't hurt anyone but calculating an average with 0 for the unknown values, you don't get the real average.

In that case, I would add price IS NOT NULL in WHERE clause to avoid these unknown values.

Answer 3

(this answer was added to provide shorter and more generic examples to the question - without including all the case-specific details in the original question).

---

There are two distinct "problems" here, the first is if a table or subquery has no rows, the second is if there are NULL values in the query.

For all versions I've tested, postgres and mysql will ignore all NULL values when averaging, and it will return NULL if there is nothing to average over. This generally makes sense, as NULL is to be considered "unknown". If you want to override this you can use coalesce (as suggested by Luc M).

$ create table foo (bar int);
CREATE TABLE

$ select avg(bar) from foo;
 avg 
-----

(1 row)

$ select coalesce(avg(bar), 0) from foo;
 coalesce 
----------
        0
(1 row)

$ insert into foo values (3);
INSERT 0 1
$ insert into foo values (9);
INSERT 0 1
$ insert into foo values (NULL);
INSERT 0 1
$ select coalesce(avg(bar), 0) from foo;
      coalesce      
--------------------
 6.0000000000000000
(1 row)

of course, "from foo" can be replaced by "from (... any complicated logic here ...) as foo"

Now, should the NULL row in the table be counted as 0? Then coalesce has to be used inside the avg call.

$ select coalesce(avg(coalesce(bar, 0)), 0) from foo;
      coalesce      
--------------------
 4.0000000000000000
(1 row)