背景
1、分兩個維度,統計數據並形成一個二維數據展現形式 2、橫軸標為統計維度1,縱軸為統計維度2 橫軸x區分為:x<=5,5<x<10,x>10 縱軸y區分為:y<=5,5<y<10,y>10 形成如下圖表格數據
| 維度2-->維度1 | x<=5 | 5<x<10 | x>10 |
|---|---|---|---|
| y<=5 | 20 | 3 | 1 |
| 5<x<=10 | 1 | 2 | 1 |
| y>10 | 0 | 0 | 1 |
步驟一:先統計出單筆兩個維度的值
select user,x,y from test01
如圖
| user | x | y |
|---|---|---|
| A | 1 | 2 |
| B | 6 | 2 |
| C | 5 | 10 |
| D | 7 | 11 |
| E | 11 | 2 |
步驟二:維度划分打標記
將每條記錄的x,y划入到對應分級中,例如: 記錄:A,x=1,y=2 划入等級為 A (x<=5) (y<=5) 記錄:B,x=6,y=2 划入等級為 A (5<x<=10) (y<=5)
select user user, (case when x<=5 then 'x<=5' when x>5 and x<=10 then '5<x<=10' when x>10 then 'x>10' else 'NULL' end) as tagX, (case when y<=5 then 'y<=5' when y>5 and y<=10 then '5<y<=10' when y>10 then 'y>10' else 'NULL' end) as tagY from test01
結果如圖:
| user | tagX | tagY |
|---|---|---|
| A | x<=5 | y<=5 |
| B | 5<x<=10 | y<=5 |
| C | x<=5 | 5<y<=10 |
| D | 5<x<=10 | y>10 |
| E | x>10 | y<=5 |
步驟三:取維度1,維度2組合-group by的結果
SQL如下
select tagX,tagY,count(1) as cnt from ( 步驟二sql )t group by tagX,tagY
即:
select tagX,tagY,count(1) as cnt from (
select user
user,
(case
when x<=5 then 'x<=5'
when x>5 and x<=10 then '5<x<=10'
when x>10 then 'x>10'
else 'NULL' end) as tagX,
(case
when y<=5 then 'y<=5'
when y>5 and y<=10 then '5<y<=10'
when y>10 then 'y>10'
else 'NULL' end) as tagY
from test01
)t
group by tagX,tagY
結果如圖:
| tagX | tagY | cnt |
|---|---|---|
| 5<x<=10 | y<=5 | 1 |
| 5<x<=10 | y>10 | 1 |
| x<=5 | 5<y<=10 | 1 |
| x<=5 | y<=5 | 1 |
| x>10 | y<=5 | 1 |
步驟四:將tagX轉成橫軸,tagY轉成縱軸
橫軸:x<=5,5<x<=10,x>10 (由於縱軸需要占用一個空間所以需要虛擬一個頂級橫軸) ---> (y/x)|x<=5|5<x<=10|x>10 縱軸:y<=5,5<y<=10,y>10 需要將tagY group by tagX通過case when 橫向平鋪
SQL如下:
select tagY as 'y/x', sum(case when tagX ='x<=5' then cnt else 0 end) as 'x<=5', sum(case when tagX ='5<x<=10' then cnt else 0 end) as '5<x<=10', sum(case when tagX ='x>10' then cnt else 0 end) as 'x>10' from( 步驟三sql )m group by tagY
即:
select tagY as 'y/x', sum(case when tagX ='x<=5' then cnt else 0 end) as 'x<=5', sum(case when tagX ='5<x<=10' then cnt else 0 end) as '5<x<=10', sum(case when tagX ='x>10' then cnt else 0 end) as 'x>10' from( select tagX,tagY,count(1) as cnt from ( select user user, (case when x<=5 then 'x<=5' when x>5 and x<=10 then '5<x<=10' when x>10 then 'x>10' else 'NULL' end) as tagX, (case when y<=5 then 'y<=5' when y>5 and y<=10 then '5<y<=10' when y>10 then 'y>10' else 'NULL' end) as tagY from test01 )t group by tagX,tagY )m group by tagY
得最終結果如圖:
| y/x | x<=5 | 5<x<=10 | x>10 |
|---|---|---|---|
| 5<y<=10 | 1 | 0 | 0 |
| y<=5 | 1 | 1 | 1 |
| y>10 | 0 | 1 | 0 |
參考:
https://blog.csdn.net/qq_25264951/article/details/56679120
https://blog.csdn.net/yoshiokayui/article/details/83564318
https://blog.csdn.net/m0_43430744/article/details/83444906
https://blog.csdn.net/John_Like_Girl/article/details/103144442
