Oracle® Database SQL Language Reference 11g Release 1 (11.1) Part Number B28286-01 |
|
|
View PDF |
Syntax
See Also:
"Analytic Functions" for information on syntax, semantics, and restrictionsPurpose
CORR
returns the coefficient of correlation of a set of number pairs. You can use it as an aggregate or analytic function.
This function takes as arguments any numeric datatype or any nonnumeric datatype that can be implicitly converted to a numeric datatype. Oracle determines the argument with the highest numeric precedence, implicitly converts the remaining arguments to that datatype, and returns that datatype.
See Also:
Table 2-10, "Implicit Type Conversion Matrix" for more information on implicit conversion and "Numeric Precedence" for information on numeric precedenceOracle Database applies the function to the set of (expr1
, expr2
) after eliminating the pairs for which either expr1
or expr2
is null. Then Oracle makes the following computation:
COVAR_POP(expr1, expr2) / (STDDEV_POP(expr1) * STDDEV_POP(expr2))
The function returns a value of type NUMBER
. If the function is applied to an empty set, then it returns null.
Note:
TheCORR
function calculates the Pearson's correlation coefficient, which requires numeric expressions as arguments. Oracle also provides the CORR_S
(Spearman's rho coefficient) and CORR_K
(Kendall's tau-b coefficient) to support nonparametric or rank correlation.See Also:
"Aggregate Functions", "About SQL Expressions" for information on valid forms ofexpr
, and CORR_* and CORR_SAggregate Example
The following example calculates the coefficient of correlation between the list prices and minimum prices of products by weight class in the sample table oe.product_information
:
SELECT weight_class, CORR(list_price, min_price) "Correlation" FROM product_information GROUP BY weight_class ORDER BY weight_class, "Correlation"; WEIGHT_CLASS Correlation ------------ ----------- 1 .999149795 2 .999022941 3 .998484472 4 .999359909 5 .999536087
Analytic Example
The following example shows the correlation between duration at the company and salary by the employee's position. The result set shows the same correlation for each employee in a given job:
SELECT employee_id, job_id, TO_CHAR((SYSDATE - hire_date) YEAR TO MONTH ) "Yrs-Mns", salary, CORR(SYSDATE-hire_date, salary) OVER(PARTITION BY job_id) AS "Correlation" FROM employees WHERE department_id in (50, 80) ORDER BY job_id, employee_id; EMPLOYEE_ID JOB_ID Yrs-Mns SALARY Correlation ----------- ---------- ------- ---------- ----------- 145 SA_MAN +08-07 14000 .912385598 146 SA_MAN +08-04 13500 .912385598 147 SA_MAN +08-02 12000 .912385598 148 SA_MAN +05-07 11000 .912385598 149 SA_MAN +05-03 10500 .912385598 150 SA_REP +08-03 10000 .80436755 151 SA_REP +08-02 9500 .80436755 152 SA_REP +07-09 9000 .80436755 153 SA_REP +07-01 8000 .80436755 154 SA_REP +06-05 7500 .80436755 155 SA_REP +05-06 7000 .80436755 ...