An aggregate function performs a calculation operation on a set of input values and returns a value. For example, the COUNT function counts the number of rows retrieved by an SQL statement. Table 1 lists aggregate functions.
Function |
Return Type |
Description |
|---|---|---|
count(*), count(expr), count(DISTINCT expr[, expr...]) |
BIGINT |
Return the total number of retrieved records. |
sum(col), sum(DISTINCT col) |
DOUBLE |
Return the sum of the values in a column. |
avg(col), avg(DISTINCT col) |
DOUBLE |
Return the average of the values in a column. |
min(col) |
DOUBLE |
Return the minimum value of a column. |
max(col) |
DOUBLE |
Return the maximum value of a column. |
variance(col), var_pop(col) |
DOUBLE |
Return the variance of a numeric column. |
var_samp(col) |
DOUBLE |
Return the sample variance of a numeric column. |
stddev_pop(col) |
DOUBLE |
Return the deviation of a numeric column. |
stddev_samp(col) |
DOUBLE |
Return the sample deviation of a numeric column. |
covar_pop(col1, col2) |
DOUBLE |
Return the covariance of a pair of numeric columns. |
covar_samp(col1, col2) |
DOUBLE |
Return the sample covariance of a pair of numeric columns. |
corr(col1, col2) |
DOUBLE |
Return the coefficient of correlation of a pair of numeric columns. |
percentile(BIGINT col, p) |
DOUBLE |
Return the exact pth percentile of a column. p must be between 0 and 1. Otherwise, this function returns null. This function does not work with floating point types. |
percentile_approx(DOUBLE col, p [, B]) |
DOUBLE |
Return an approximate pth percentile of a numeric column (including floating point types) in a group. p must be between 0 and 1. B controls approximation accuracy. Higher values of B mean better approximations, and the default value is 10,000. When the number of distinct values in the numeric column is smaller than B, an exact percentile value is returned. |
Functions such as var_pop, stddev_pop, var_samp, stddev_samp, covar_pop, covar_samp, corr, and percentile_approx, do not support non-numeral data types, such as TimeStamp.