100 important knowledge points that SQL must master: grouping data

10.1 The data packet
I learned from the last lesson , Use SQL Aggregate functions can aggregate data . such , We can
Count rows , Calculate the sum and average , Get the maximum and minimum values without retrieving all the data
Small value .
All calculations so far have been based on all data in the table or matching a specific WHERE Number of clauses
According to the above . For example, the following example returns the supplier DLL01 Number of products offered ：
Input ▼
SELECT COUNT(*) AS num_prods
FROM Products
WHERE vend_id = 'DLL01';
Output ▼
num_prods
-----------
4

If you want to return the number of products provided by each supplier , What should I do ？ Or return with only one
The product of the supplier of the product , Or return to provide 10 Products from suppliers of more than products ,
What do I do ？
This is the time for groups to show their skills . Using grouping, you can divide data into multiple logical groups ,
Aggregate each group .
10.2 Create grouping
Grouping is using SELECT Of the statement GROUP BY Clause . The best way to understand grouping
Look at an example ：
Input ▼
SELECT vend_id, COUNT(*) AS num_prods
FROM Products
GROUP BY vend_id;
Output ▼
vend_id num_prods
------- ---------
BRS01 3
DLL01 4
FNG01 2
analysis ▼
above SELECT Statement specifies two columns ： vend_id Include product suppliers ID,
num_prods For the calculated field （ use COUNT(*) Function creation ）. GROUP BY Clause indicates
DBMS Press vend_id Sort and group data . This will be for everyone vend_id Not the whole thing
Table calculation num_prods once . As you can see from the output , supplier BRS01 Yes 3 Individual property
product , supplier DLL01 Yes 4 A product , And suppliers FNG01 Yes 2 A product .

Because of the use of GROUP BY , It is not necessary to specify each group to be calculated and valued . The system will automatically
It's done . GROUP BY Clause indicates DBMS Grouped data , Then for each group instead of the whole
Result sets are aggregated .
In the use of GROUP BY Before clause , You need to know some important rules .
 GROUP BY Clause can contain any number of columns , Thus, groups can be nested ,
More detailed data grouping .
   If in GROUP BY Group nested in Clause , The data will advance on the last specified group
Row summary . let me put it another way , When creating a group , All columns specified are calculated together （ therefore
Data cannot be retrieved from individual columns ）.
 GROUP BY Each column listed in the clause must be a retrieval column or a valid expression （ but
Cannot be an aggregate function ）. If in SELECT Using expressions in , Must be in GROUP BY
Clause specifies the same expression . Cannot use alias .
   majority SQL Implementation does not allow GROUP BY Columns have variable length data types （ Such as Wen
This or remark type field ）.
   Except for aggregate calculation statements , SELECT Each column in the statement must be in GROUP BY Clause
Give in .
   If the grouping column contains NULL Row of values , be NULL Will be returned as a group .
If there are more than one row in the column NULL value , They will be divided into groups .
 GROUP BY Clause must appear in WHERE After Clause , ORDER BY Before clause .

Tips ： ALL Clause
Microsoft SQL Server Wait some SQL Realize in GROUP BY Optional... Is supported in ALL
Clause . This clause can be used to return all groups , Even groups that do not have matching rows return
return （ In this case , The aggregation will return NULL ）. Concrete DBMS Do you support ALL ,
Please refer to the corresponding documentation .

Be careful ： Specify columns by relative position
yes , we have SQL The implementation allows you to SELECT The position in the list specifies GROUP BY The column of .
for example , GROUP BY 2, 1 Can represent grouping by the second column selected , Then press the first
Columns grouped . Although this shorthand grammar is very convenient , But not all SQL The implementation supports ,
And it's easy to edit SQL Error in statement  

10.3 Filter grouping
Apart from being able to use GROUP BY Out of packet data ,SQL It also allows filtering groups , What does the regulation include
grouping , Which groups are excluded . for example , You may want to list all customers who have at least two orders
customer . So , Filtering must be based on complete grouping rather than individual rows .
We have seen WHERE The function of clause （ The first 4 Class mention ）. however , In this example
in WHERE Unable to complete the task , because WHERE Filter specifies rows rather than groups . The facts
On , WHERE No concept of grouping .
that , Don't use WHERE What to use ？SQL Another clause is provided for this purpose , Namely
HAVING Clause . HAVING Very similar to WHERE . in fact , What I have learned so far
All types of WHERE Clause can be used HAVING To replace . The only difference is , WHERE
Filter line , and HAVING Filter grouping .
Tips ： HAVING Support all WHERE The operator
In the 4 Lesson and lesson 5 In class , We learned WHERE Condition of clause （ Including wildcard bar
And clauses with multiple operators ）. What I've learned is about WHERE All technologies and options
Items apply to HAVING . Their syntax is the same , Only the key words are different .
that , How to filter groups ？ Look at the following example ：

Input ▼
SELECT cust_id, COUNT(*) AS orders
FROM Orders
GROUP BY cust_id
HAVING COUNT(*) >= 2;
Output ▼
cust_id orders
---------- -----------
1000000001 2
analysis ▼
This article SELECT The first three lines of the statement are similar to the above statement . The last line adds HAVING
Clause , It filters COUNT(*) >= 2 （ More than two orders ） Those groupings .
You can see , WHERE Clause doesn't work here , Because filtering is based on grouping aggregation values ,
Not the value of a particular row .

explain ： HAVING and WHERE The difference between
Here is another way to understand , WHERE Filter before data grouping , HAVING In number
Filter after grouping . This is an important difference , WHERE Excluded lines are not included in
In group . This may change the calculated value , Thereby affecting HAVING Clause based on these values
Filtered groups .

that , Have you used... In one statement at the same time WHERE and HAVING The need for clauses ？ The facts
On , It does . If you want to further filter the above statements , Return it to the past 12 Within months
Customers with more than two orders . So , You can add one WHERE Clause , Filter out the past 12 Months
Orders placed within , And then add HAVING Clause to filter out groups with more than two orders .
Just to understand , Let's look at the following example , It lists two or more products and their prices
Greater than or equal to 4 Supplier ： 

Input ▼
SELECT vend_id, COUNT(*) AS num_prods
FROM Products
WHERE prod_price >= 4
GROUP BY vend_id
HAVING COUNT(*) >= 2;
Output ▼
vend_id num_prods
------- -----------
BRS01 3
FNG01 2
analysis ▼
In this statement , The first line is the basic... Using the aggregation function SELECT sentence , It's very similar to the previous one
Example . WHERE Clause filter all prod_price At least for 4 The line of , Then press vend_id
Grouped data , HAVING Clause filter count is 2 or 2 Groups above . without WHERE
Clause , One more line will be retrieved （ supplier DLL01 , sales 4 A product , The prices are all in 4
following ）：
Input ▼
SELECT vend_id, COUNT(*) AS num_prods
FROM Products
GROUP BY vend_id
HAVING COUNT(*) >= 2;
Output ▼
vend_id num_prods
------- -----------
BRS01 3
DLL01 4
FNG01 2

explain ： Use HAVING and WHERE
HAVING And WHERE Very similar , If you don't specify GROUP BY , Most of them DBMS
Will treat them equally . however , You have to be able to distinguish this for yourself . Use HAVING Should be
The combination GROUP BY Clause , and WHERE Clause is used for standard row level filtering . 

10.4 Grouping and sorting
GROUP BY and ORDER BY Do the same job often , But they are very different , Understand this
This is very important . surface 10-1 Sum up the differences between them .
surface 10-1  ORDER BY And GROUP BY 
ORDER BY  GROUP BY
Sort the resulting output   Group lines , But the output may not be in the order of grouping
Any column can use （ Not even
The selected column can also use ）
Only selection columns or expression columns can be used , And you must use each selection column
expression
Not necessarily   If you use columns with aggregate functions （ Or expressions ）, Must be used
surface 10-1 The first difference listed in is extremely important . We often find , use GROUP BY grouping
The data of is indeed output in grouping order . But not always , This is not SQL Specifications
Required . Besides , Even if a particular DBMS Always follow the given GROUP BY Number of clause sorts
According to the , Users may also require sorting in different order . Just because you group numbers in some way
According to the （ Get a specific group aggregation value ）, That doesn't mean you need to sort the output the same way .
Should provide clear ORDER BY Clause , Even if the effect is equivalent to GROUP BY Clause .

Tips ： Don't forget it ORDER BY
Generally in use GROUP BY When clause , It should also be given ORDER BY Clause . This is Bao
The only way to verify that the data is sorted correctly . Don't just rely on GROUP BY Sorting data . 

To illustrate GROUP BY and ORDER BY How to use , Let's look at an example . Below SELECT

The statement is similar to the previous examples . It retrieves order numbers and orders that contain three or more items
The number of items ：
Input ▼
SELECT order_num, COUNT(*) AS items
FROM OrderItems
GROUP BY order_num
HAVING COUNT(*) >= 3;
Output ▼
order_num items
--------- -----
20006 3
20007 5
20008 5
20009 3
To sort the output by the number of items ordered , Need to add ORDER BY Clause , As shown below ：
Input ▼
SELECT order_num, COUNT(*) AS items
FROM OrderItems
GROUP BY order_num
HAVING COUNT(*) >= 3
ORDER BY items, order_num;
Output ▼
order_num items
--------- -----
20006 3
20009 3
20007 5
20008 5

analysis ▼
In this case , Use GROUP BY Clause by order number （ order_num Column ） Number of groups
According to the , In order to COUNT(*) Function can return the number of items in each order . HAVING Clause
Filtering data , So that only orders containing three or more items are returned . Last , use ORDER BY
Clause sort output .
10.5  SELECT Clause order
Let's review SELECT The order of sentences in a sentence . surface 10-2 In the SELECT In the sentence
The order that must be followed when using , List the clauses learned so far .
surface 10-2  SELECT Clauses and their order
Son sentence   say bright   Whether it is necessary to use
SELECT
Column or expression to return   yes
FROM
A table from which to retrieve data   Use only when selecting data from a table
WHERE
Row level filtering   no
GROUP BY
Group description   Use only when aggregating by group
HAVING
Group level filtering   no
ORDER BY
Output sort order   no