在 Oracle 中使用 SQL 计算相关矩阵
Calculate correlation matrix using SQL in Oracle
给定样本 table 具有如下值：
a1
a2
a3
a4
a5
12
15
13
12
21
14
23
41
57
13
23
41
57
13
39
90
81
32
76
78
我们的目标是在 Oracle 中使用 SQL 创建相关矩阵。
结果table应该是这样的：
#
a1
a2
a3
a4
a5
a1
1
0.968627486
-0.023590701
0.723566409
0.962065691
a2
0.968627486
1
0.214217652
0.655979092
0.974931214
a3
-0.023590701
0.214217652
1
0.009891122
0.079502115
a4
0.723566409
0.655979092
0.009891122
1
0.524414529
a5
0.962065691
0.974931214
0.079502115
0.524414529
1
我尝试了一些公式来计算两个值的相关系数，但没有用。相关函数也不起作用。
您可以使用 CORR function:
WITH ordered (rn, a1, a2, a3, a4, a5) AS (
SELECT ROWNUM, a1, a2, a3, a4, a5 FROM table_name
)
SELECT key,
CORR(a1, value) AS a1,
CORR(a2, value) AS a2,
CORR(a3, value) AS a3,
CORR(a4, value) AS a4,
CORR(a5, value) AS a5
FROM ordered t
INNER JOIN (
SELECT *
FROM ordered
UNPIVOT (
value FOR key IN (a1, a2, a3, a4, a5)
)
) d
ON (t.rn = d.rn)
GROUP BY key
其中，对于示例数据：
CREATE TABLE table_name (a1, a2, a3, a4, a5) AS
SELECT 12, 15, 13, 12, 21 FROM DUAL UNION ALL
SELECT 14, 23, 41, 57, 13 FROM DUAL UNION ALL
SELECT 23, 41, 57, 13, 39 FROM DUAL UNION ALL
SELECT 90, 81, 32, 76, 78 FROM DUAL;
输出：
KEY
A1
A2
A3
A4
A5
A1
1
.9686274863493618022648973908306711860158
-.0235907009791586343670275607353945975836
.7235664093128934907375377255687028401814
.9620656911630149496690023500551903300058
A2
.9686274863493618022648973908306711860158
1
.2142176524658728558977962209541976877097
.655979092200627387045632202651644194262
.9749312135706365317565525851529784515689
A3
-.0235907009791586343670275607353945975836
.2142176524658728558977962209541976877097
1
.009891122282267634886059354456616140169476
.079502115405919407497390073647822157651
A4
.7235664093128934907375377255687028401814
.655979092200627387045632202651644194262
.009891122282267634886059354456616140169476
1
.5244145293486588849105425925771231892634
A5
.9620656911630149496690023500551903300058
.9749312135706365317565525851529784515689
.079502115405919407497390073647822157651
.5244145293486588849105425925771231892634
1
db<>fiddle here
给定样本 table 具有如下值：
| a1 | a2 | a3 | a4 | a5 |
|---|---|---|---|---|
| 12 | 15 | 13 | 12 | 21 |
| 14 | 23 | 41 | 57 | 13 |
| 23 | 41 | 57 | 13 | 39 |
| 90 | 81 | 32 | 76 | 78 |
我们的目标是在 Oracle 中使用 SQL 创建相关矩阵。
结果table应该是这样的：
| # | a1 | a2 | a3 | a4 | a5 |
|---|---|---|---|---|---|
| a1 | 1 | 0.968627486 | -0.023590701 | 0.723566409 | 0.962065691 |
| a2 | 0.968627486 | 1 | 0.214217652 | 0.655979092 | 0.974931214 |
| a3 | -0.023590701 | 0.214217652 | 1 | 0.009891122 | 0.079502115 |
| a4 | 0.723566409 | 0.655979092 | 0.009891122 | 1 | 0.524414529 |
| a5 | 0.962065691 | 0.974931214 | 0.079502115 | 0.524414529 | 1 |
我尝试了一些公式来计算两个值的相关系数，但没有用。相关函数也不起作用。
您可以使用 CORR function:
WITH ordered (rn, a1, a2, a3, a4, a5) AS (
SELECT ROWNUM, a1, a2, a3, a4, a5 FROM table_name
)
SELECT key,
CORR(a1, value) AS a1,
CORR(a2, value) AS a2,
CORR(a3, value) AS a3,
CORR(a4, value) AS a4,
CORR(a5, value) AS a5
FROM ordered t
INNER JOIN (
SELECT *
FROM ordered
UNPIVOT (
value FOR key IN (a1, a2, a3, a4, a5)
)
) d
ON (t.rn = d.rn)
GROUP BY key
其中，对于示例数据：
CREATE TABLE table_name (a1, a2, a3, a4, a5) AS
SELECT 12, 15, 13, 12, 21 FROM DUAL UNION ALL
SELECT 14, 23, 41, 57, 13 FROM DUAL UNION ALL
SELECT 23, 41, 57, 13, 39 FROM DUAL UNION ALL
SELECT 90, 81, 32, 76, 78 FROM DUAL;
输出：
KEY A1 A2 A3 A4 A5 A1 1 .9686274863493618022648973908306711860158 -.0235907009791586343670275607353945975836 .7235664093128934907375377255687028401814 .9620656911630149496690023500551903300058 A2 .9686274863493618022648973908306711860158 1 .2142176524658728558977962209541976877097 .655979092200627387045632202651644194262 .9749312135706365317565525851529784515689 A3 -.0235907009791586343670275607353945975836 .2142176524658728558977962209541976877097 1 .009891122282267634886059354456616140169476 .079502115405919407497390073647822157651 A4 .7235664093128934907375377255687028401814 .655979092200627387045632202651644194262 .009891122282267634886059354456616140169476 1 .5244145293486588849105425925771231892634 A5 .9620656911630149496690023500551903300058 .9749312135706365317565525851529784515689 .079502115405919407497390073647822157651 .5244145293486588849105425925771231892634 1
db<>fiddle here