Nice work! There's one more thing about GROUP BY
that we want to discuss. Sometimes we want to group the rows by more than one column. Let's imagine we have a few customers who place tons of orders every day, so we would like to know the daily sum of their orders.
SELECT
customer_id,
order_date,
SUM(total_sum)
FROM orders
GROUP BY customer_id, order_date;
As you can see, we group by two columns: customer_id
and order_date
. We select these columns along with the function SUM(total_sum)
.
Remember: in such queries each column in the SELECT
part must either be used later for grouping or it must be used with one of the functions.
To better understand the issue, take a look at1 the following table (expand the column or scroll the table horizontally if you need to):
order_id |
customer_id |
order_date |
ship_date |
total_sum |
|
customer_id |
order_date |
SUM(total_sum) |
16 |
6 |
2015-03-28 |
2015-03-29 |
230.54 |
|
6 |
2015-03-28 |
2906.19 |
17 |
6 |
2015-03-28 |
2015-03-30 |
321.42 |
|
18 |
6 |
2015-03-28 |
2015-03-30 |
2354.23 |
|
19 |
6 |
2015-03-29 |
2015-03-30 |
4224.21 |
|
6 |
2015-03-29 |
10788.06 |
20 |
6 |
2015-03-29 |
2015-03-30 |
2314.32 |
|
21 |
6 |
2015-03-29 |
2015-03-31 |
124.21 |
|
22 |
6 |
2015-03-29 |
2015-03-31 |
4125.32 |
|
23 |
6 |
2015-03-30 |
2015-04-03 |
645.23 |
→ |
6 |
2015-03-30 |
10504.42 |
24 |
6 |
2015-03-30 |
2015-04-05 |
7543.56 |
|
25 |
6 |
2015-03-30 |
2015-04-05 |
2315.63 |
|
26 |
7 |
2015-04-02 |
2015-04-05 |
523.98 |
|
7 |
2015-04-02 |
9580.42 |
27 |
7 |
2015-04-02 |
2015-04-06 |
523.13 |
|
28 |
7 |
2015-04-02 |
2015-04-07 |
8533.31 |
|
29 |
7 |
2015-04-03 |
2015-04-07 |
4245.64 |
|
7 |
2015-04-03 |
4245.64 |
Note: It makes no sense to select any other column. For example, each order on the very same day by the very same customer can have a different shipping date. If you wanted to select the column ship_date
in this case, the database wouldn't know which shipping date to choose for the whole group, so it would put just one, random value in the result.