Question

Been trying to sort this one out for a while. I’d really appreciate any help.

I’ve got this table where I’m getting 2 columns with date and int values respectively. The problem is that mysql skips the date values wherever the int value is null.

Here the sql statement

SELECT DATE_FORMAT(sales_date_sold, '%b \'%y')
    AS sale_date, sales_amount_sold 
    AS sale_amt 
    FROM yearly_sales 
    WHERE sales_date_sold BETWEEN DATE_SUB(SYSDATE(), INTERVAL 2 YEAR) AND SYSDATE() 
    GROUP BY YEAR(sales_date_sold), MONTH(sales_date_sold) 
    ORDER BY YEAR(sales_date_sold), MONTH(sales_date_sold) ASC;

There aren’t any values for feb 2011 so that month gets skipped, along with a few others. Coalesce and if_null don’t work too.

Answer 1

You need a row source that provides values for all of the months in the dimension, and then left join your yearly_sales table to that.

You are doing a GROUP BY, you most likely want an aggregate on your measure (sales_amount_sold), or you don’t want a GROUP BY. (The query in your question is going to return a value from sales_amount_sold for only one row in a given month. That may be what you want, but its a very odd resultset to return.)

---

One approach is to have a “calendar_month” table that contains DATE values all of the months you want returned. (There are other ways to generate this, existing answers to questions elsewhere on stackoverflow)

SELECT m.month                            AS sale_date
     , IFNULL(SUM(s.sales_amount_sold),0) AS sale_amt
  FROM calendar_months m
  LEFT
  JOIN yearly_sales s
    ON s.sales_date_sold >= m.month 
   AND s.sales_date_sold < DATE_ADD(m.month,INTERVAL 1 MONTH)
 WHERE m.month BETWEEN DATE_SUB(SYSDATE(), INTERVAL 2 YEAR) AND SYSDATE()
 GROUP BY m.month
 ORDER BY m.month

This query returns a slightly different result, you are only going to get rows in groups of “whole months”, rather than including partial months, as in your original query, because the WHERE clause on sale_date references two years before the current date and time, rather than the “first of the month” two years before.

A calendar_months table is not necessarily required; this could be replaced with a query that returns the row source. In that case, the predicate on the month value could be moved from the outer query into the subquery.

---

Addendum: if you use a calendar_month table as a rowsource, you’ll need to populate it with every possible “month” value you want to return.

CREATE TABLE calendar_month
(`month` DATE NOT NULL PRIMARY KEY COMMENT 'one row for first day of each month');

INSERT INTO calendar_month(`month`) VALUES ('2011-01-01'),('2011-02-01'),('2011-03-01') 

---

As an alternative, you can specify a dynamically generated rowsource, as an inline view, rather than a table reference. (You could use a similar query to quickly populate a calendar_months table.)

You can wrap this query in parenthesis, and paste it between FROM and calendar_months in the previous query I provided.

SELECT DATE_ADD('1990-01-01',INTERVAL 1000*thousands.digit + 100*hundreds.digit + 10*tens.digit + ones.digit MONTH) AS `month`
  FROM ( SELECT 0 AS digit UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3 UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6 UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9 ) ones
  JOIN ( SELECT 0 AS digit UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3 UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6 UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9 ) tens
  JOIN ( SELECT 0 AS digit UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3 UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6 UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9 ) hundreds
  JOIN ( SELECT 0 AS digit UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3 UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6 UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9 ) thousands