A few ways to do a SQL

One day in a technology group to see someone sent such a figure for help, a look at the problem is to examine the use of SQL row to column

I thought the first time I think of Oracle's decode function to write, but also think that the topic is not specifically what kind of database ~ If you use decode, under MySQL is completely different usage (also because this time I know that MySQL also has decode method, But unlike Oracle, other databases are not being researched.

So I responded by writing a SQL

The first thought of sql¶

SelectT1.d,t1.c1'win', T2.C2'Negative'  from(Select Count(Result) c1,d fromT_resultwhereResult= 'win' Group  byd) T1 Left outer Join(Select Count(Result) c2,d fromT_resultwhereResult= 'Negative' Group  byd) T2 onT1.d=T2.d

This kind of writing is more intuitive, is to take the "win" out, in taking "negative" out, and finally associated with the date

PS: Here the T_result is the data table, D is the time column, result is the outcome column

I thought about it later, so the result is out, but in fact, the efficiency is not high, because two times select will lead to two of the table scan, because the previous study of Oracle, these are still more sensitive to these, so later rewritten a bit

(May not write the best, but at least more than the initial progress, performance has improved, after all, optimization is a step-by-step)

SELECTDSUM( Case  whenResult='win'  Then 1             ELSE 0             END),         SUM( Case  whenResult='Negative'  Then 1             ELSE 0             END) fromT_resultGROUP  byD

So sweep the surface once and the table will be able to produce results.

Finally, we enclose the decode function of Oralce.

SELECT D,sum(decode (result,' win ',1,0)),sum (Decode (result,' negative ',1,0)) from T_result GROUP  by D

A few ways to do a SQL