How to achieve following result with Group By in MySql? - mysql

Question

Consider following table:
+------------+-----------+-------------+---------+------------+--------+-------------+
| client_id | TradeDate | servicetype | SEGMENT | OrdChannel | orders | OrderAmount |
+------------+-----------+-------------+---------+------------+--------+-------------+
| 1 | 20140611 | Type_1 | CASH | TT | 1 | 39275 |
| 2 | 20150119 | Type_1 | CASH | DNT | 2 | 11856.9 |
| 3 | 20150922 | Type_1 | FNO | OTHER | 1 | 854750 |
| 4 | 20151223 | Type_1 | CASH | TT | 5 | 71075 |
| 5 | 20140529 | Type_1 | Offline | FNO | 1 | 0 |
| 6 | 20160310 | Type_2 | CASH | WEB | 2 | 8009.6 |
| 7 | 20150318 | Type_1 | Offline | FNO | 2 | 432900 |
| 8 | 20150914 | Type_2 | CASH | WEB | 2 | 15612 |
| 9 | 20160317 | Type_2 | FNO | MINI | 1 | 9000 |
| 10 | 20140421 | Type_1 | CASH | TT | 8 | 17112.5 |
+------------+-----------+-------------+---------+------------+--------+-------------+
I am trying to group this data by client_id and TradeDate. Thus final dataset would contain each row for every <client-id, TradeDate> pair.
I want to compute following features from this data:
For each type of SEGMENT in every pair, I want to compute sum of orders and OrderAmount
Similarly, for each type of OrdChannel in every pair, I want to compute sum of orders and OrderAmount
Finally, a count of each servicetype i.e. Type_1 and Type_2.
Thus final dataset would contain columns similar to shown below:
+------------+-----------+-------------+---------+------------+--------+-------------+-------------+-------------+-------------+-------------+-------------+-------------+-------------+-------------+-------------+
| client_id | TradeDate | SEGMENT_CASH_orders_sum | SEGMENT_CASH_OrderAmount_sum.... | OrdChannel_TT_orders_sum | OrdChannel_TT_OrderAmount_sum.... | servicetype_Type_1_count | servicetype_Type_2_count |
+------------+-----------+-------------+---------+------------+--------+-------------+-------------+-------------+-------------+-------------+-------------+-------------+-------------+-------------+-------------+
So far, I have tried:
select clientsrno, TradeDate, SEGMENT, COUNT(orders) from orders group by clientsrno, TradeDate, SEGMENT;
but it's not giving separate columns for SEGMENT_CASH-orders_sum, etc.
Instead the output I am getting is:
+------------+-----------+---------+---------------+
| clientsrno | TradeDate | SEGMENT | COUNT(orders) |
+------------+-----------+---------+---------------+
| 44 | 20141209 | CASH | 23 |
| 44 | 20141211 | FNO | 10 |
+------------+-----------+---------+---------------+

Answer 1

You can do with a lot of sum( CASE .... END)
this is a suggestionfor the firts column
select client_di, TradeDate
sum (case SEGMENT
when 'CASH' then orders ELSE 0 END) as SEGMENT_CASH_orders_sum,
sum (case SEGMENT
when 'CASH' then OrderAmount ELSE 0 END) as SEGMENT_CASH_ordersAmount_sum,
sum (case SEGMENT
when 'FNO' then orders ELSE 0 END) as SEGMENT_FNO_orders,
sum (case SEGMENT
when 'FNO' then OrderAmount ELSE 0 END) as SEGMENT_FNO_ordersAmount_sum
from my_table
group by client_id, TradeDate

Answer 2

You can do for example:
SELECT clientsrno, TradeDate, SEGMENT, COUNT(orders), sumOrders.total FROM tabletest
INNER JOIN (
SELECT tabletest.tradeDate, SUM(orders) as total FROM tabletest
GROUP BY clientsrno, TradeDate, SEGMENT
) sumOrders ON tableTest.tradeDate = sumOrders.tradeDate
GROUP BY clientsrno, TradeDate, SEGMENT
And repeat the process for every sum you need.