Question

I have a table with sometable:

name |test_date | score   
-----------------------
jon  |2012:07:01|   95   
jon  |2012:07:01|   60   
jon  |2012:07:01|   30   
alex |2012:07:01|   80   
alex |2012:07:01|   85   
alex |2011:05:01|   40  
emile|2011:01:01|   89    

I want to get some rows for each name and give information about his score_grading with rules:
score>79=A,
80>score>49=B
otherwise C.

The problem is, I want: If in the same day there are more than one same score_grade for a student, than it will count as one score_grade.
For example, in the table we can see that alex get A two times in one day, I want it to be counted as only 1 A.

So the result will be

name |    A   |    B     |    C
jon  |    1   |    1     |    1
alex |    1   |    0     |    1  
emily|    1   |    0     |    0

I only know code like:

SELECT name,
SELECT SUM(IF(score)>79),1,0)) as A,
SELECT SUM(IF(80>score>49),1,0)) as B,
SELECT SUM(IF(score)<50),1,0)) as C from sometable group by name

Nha, how do I put a “DISTINCT” on it?
Can anybody give a solution? Maybe it does not need a DISTINCT?
Thanks. ^^

Answer 1

You mean something like the following (sqlfiddle)?

SELECT name, SUM(IF(A>=1,1,0)) as A, SUM(IF(B>=1,1,0)) as B, 
    SUM(IF(C>=1,1,0)) as C
FROM
(
    SELECT name, test_date, SUM(IF(score>79,1,0)) as A, 
        SUM(IF(score BETWEEN 49 AND 79,1,0)) as B, SUM(IF(score<50,1,0)) as C 
    FROM sometable 
    GROUP BY name, test_date
) daygroups
GROUP BY name

This first shuffles your data into rows of name, test_date, A, B, C in the subquery. Then the outer query will aggregate these rows, taking a 1 if there is at least one score of that letter grade on the day, else taking a 0.

---

This ought to work too (sqlfiddle):

SELECT name, SUM(IF(lettergrade = 'A',1,0)) AS A, 
    SUM(IF(lettergrade = 'B',1,0)) AS B, SUM(IF(lettergrade = 'C',1,0)) AS C
FROM
(
    SELECT DISTINCT name, test_date, 
        CASE WHEN score>79 THEN 'A' 
             WHEN score BETWEEN 49 AND 79 THEN 'B' 
             ELSE 'C' 
        END AS lettergrade
    FROM sometable
) lettergrades
GROUP BY name

I’m not sure which one would be better. This query uses a DISTINCT as you suggest in your question. First it resolves each numeric grade into the corresponding letter grade, then DISTINCTs out the duplicates. Finally it shuffles the data into columns.