1. A good comprehension model:
1) Where expression: Fit this experssion into every row and get the true/false result.
#Where expression comprehension model select * from stu where stu_name = 'zhangsan';
2) Row: Regard as variable which enables basic math operation.
#Row comprehension model #We can regard the second 'stu_name' as a variable whose value depends on the row cursor now pointing to select stu_name from stu where stu_name = 'zhangsan'
3) Result set: Regard as a temporary table which enables further query.
#We regard the sub-query result set as a temp table which enables further query #We want to find the stu_name whose max score equals or larger than 80 select * from (select stu_name, max(stu_score) as stu_max from stu group by stu_name) as temp where stu_max >= 80;
2. Four Types of Sub-query
1) Where sub-query
# Find the highest score and the corresponding stu_name, stu_course # Solution 1: Where sub-query # Regard the result of where sub-query as a value select stu_name, stu_score, stu_course from stu where stu_score = (select max(stu_score) from stu); # Solution 2: Order & Limit select stu_name, stu_score, stu_course from stu group by stu_score limit 1;
#Find the highest score and the corresponding student in each course #Using where sub-query #But this kind of approach is not that precise. #Just use the stu_score as the signal, there may be some other stu_score in other courses that equals the max(score). select * from stu where stu_score in (select max(stu_score) from stu group by stu_course); #There is an overlap in Literature course because the stu_score in Literature 50 equals the highest score that in Geograph 50 +----------+------------+-----------+ | stu_name | stu_course | stu_score | +----------+------------+-----------+ | zhangsan | Math | 90 | | zhangsan | Literature | 50 | | lisi | Literature | 55 | | zhaoliu | Geograph | 50 | | zhaoliu | Politic | 99 | +----------+------------+-----------+
2) In sub-query
#The core thought lies behind From sub-query is regarding the result of inner query as a table based on which we can start a new outter query. select * from (select max(stu_score) from stu group by stu_course) as temp;
#Find the avg score for students who has more than one invalid course. #Using Where sub-query #Most inner query find the stu_name who has more than one invalid course. #Secondary inner query filter other useless columns for where sub-query #Outter query find the average score for each students whose name lists in inner query. select stu_name, avg(stu_score) from stu where stu_name in (select stu_name from (select stu_name, stu_score, sum(stu_score < 60) as invalid_count from stu group by stu_name having invalid_count >= 2) as temp) group by stu_name; #Another way to find out the student name whose invalid course in more than one select stu_name, count(*) as invalid_count from stu where stu_score < 60 group by stu_name having invalid_count >= 2;
3) Exists sub-query
#Core thoughts lies on Exists sub-query is that 'Put the result of out query into inner query to judge the inner query true or false' select * from category; #First Level Category ####Secondary Category #######Real Goods #######Real Goods ####Secondary Category #######Third Level Category ###########Real Goods ####Secondary Category # All the categories are stored in category table, # But some categories contains no goods but secondary category # What if we want to figure out the category that directly contains goods? select cat_id, cat_name from category where exists(select * from goods where goods.cat_id = category.cat_id);