如何将一个SQL查询的结果传递给另一个查询

Question

我正在尝试使用以下 SQL 查询检索包含一组字段上的 Cramer 相关系数值的表：

WITH var_pairs AS (
        WITH vars AS (SELECT n FROM unnest(ARRAY['Performance Score', 'state', 'sex', 'maritaldesc', 'citizendesc', 'Hispanic/Latino', 'racedesc', 'Reason For Term', 'Employment Status', 'department', 'position', 'Manager Name', 'Employee Source']) AS n)
        SELECT vars1.n AS var1, vars2.n AS var2
        FROM vars AS vars1 CROSS JOIN vars AS vars2
    )
    SELECT
        (WITH
            -- Contingency table
            observed AS (
                SELECT 
                    var1 AS x, 
                    var2 AS y, 
                    COUNT(*) AS observed
                FROM hr_dataset
                GROUP BY var1, var2
            ),
            -- Sum of the rows of the contingency table
            row_total AS (
                SELECT 
                    x,
                    SUM(observed) AS row_total
                FROM observed
                GROUP BY x
            ),
            -- Sum of columns of the contiguity table
            col_total AS (
                SELECT
                    y,
                    SUM(observed) AS col_total
                FROM observed
                GROUP BY y
            ),
            -- Total number of observations
            grand_total AS (
                SELECT SUM(observed) AS grand_total
                FROM observed
            ),
            -- Expected frequencies
            expected AS (
                SELECT
                    observed.x,
                    observed.y,
                    (row_total.row_total * col_total.col_total) / grand_total.grand_total AS expected
                FROM
                    observed
                    INNER JOIN row_total USING(x)
                    INNER JOIN col_total USING(y)
                    CROSS JOIN grand_total
            ),
            -- Chi-square statistics
            chi_sq AS (
                SELECT
                    SUM(POWER(observed.observed - expected.expected, 2) / expected.expected) AS chi_sq
                FROM
                    observed 
                    INNER JOIN expected USING(x, y)
            ),
            count_x AS (
                SELECT 
                    count(DISTINCT x) AS count_x
                FROM
                    observed
            ),
            count_y AS (
                SELECT 
                    count(DISTINCT y) AS count_y
                FROM
                    observed
            )
        -- Cramer's correlation coefficient
        SELECT 
            SQRT(chi_sq / (grand_total * (least(count_x, count_y) - 1))) AS "Cramer\'s V"
        FROM 
            chi_sq
            CROSS JOIN grand_total
            CROSS JOIN count_x
            CROSS JOIN count_y
    )
    FROM var_pairs

服务器响应：错误：除以零。据我了解，我错误地将参数 var1 和 var2 传递给子查询。怎样做才正确呢？我正在使用 Postgres。

Answer 1

不，这是除以零的错误，这意味着您在某些时候尝试执行此模式

a / b

其中

b

是 0。在数学中除以零是一个特殊问题。

假设您除以 a / 0，并假设其结果是 r：

a / 0 = r

现在，让我们将方程乘以 0，得到：

a = 0 * r

那么，对于哪个 r 来说，a = 0 * r 是正确的？答案是，只要 a 为 0，它对于任何 r 都是正确的。但是，只要 a 不为 0，它对于任何 r 都是错误的。

因此：

a / 0

如果 a <> 0 和 a / 0 可以是任意值（如果 a = 0），则毫无意义。

因此，除以零是数学中的一个特例，编程语言可以处理这个问题。

由于您的分母不依赖于

var1

或

var2

的值，因此我们讨论的问题完全独立于您如何传递它们，问题是您除以零。要解决您的问题，您需要确保不除以零，因此您检查分母是否为零，如果是，则在其中设置默认值，否则进行除法。比如：

case
    when denominator = 0 then 100
    else numerator / denominator
end