Question

我有一个名为LOGENTRY的简单表，其中包含名为“DATE”和“COST”的字段。例如：

+--------------+-------+
| DATE         | COST  |
+--------------+-------+
| MAY 1 2013   | 0.8   |
| SEP 1 2013   | 0.4   |
| NOV 1 2013   | 0.6   |
| DEC 1 2013   | 0.2   |
+--------------+-------+

我想找到COST字段随时间的斜率（选择的行数范围），得到 SLOPE = -0.00216（这相当于Excel的SLOPE函数，即线性回归）。

有没有一种简单的方法来选择COST的斜率？如果我用调用语言（php）进行数学计算，我可以找到斜率：

SLOPE =  (N * Sum_XY - Sum_X * Sum_Y)/(N * Sum_X2 - Sum_X * Sum_X);

我看到一些类似的问题，但它们更复杂。我试图将这个例子剥离到最简单的情况 - 所以我能理解答案:)这里和我一样接近......但是MYSQL抱怨附近的语法： 'float））AS Sum_X，SUM（CAST（LOGENTRY.DATE as float）* CAST（LOGENTRY.DATE'

SELECT 
  COUNT( * ) AS N, 
  SUM( CAST( LOGENTRY.DATE AS FLOAT ) ) AS Sum_X, 
  SUM( CAST( LOGENTRY.DATE AS FLOAT ) * CAST( LOGENTRY.DATE AS FLOAT ) ) AS Sum_X2, 
  SUM( LOGENTRY.COST ) AS Sum_Y, SUM( LOGENTRY.COST * LOGENTRY.COST ) AS Sum_Y2, 
  SUM( CAST( LOGENTRY.DATE AS FLOAT ) * LOGENTRY.COST ) AS Sum_XY
FROM LOGENTRY

Answer 1

似乎MySQL无法将日期转换为float（根据stackoverflow中的其他示例）。也许其他示例引用另一个数据库。因此，通过将日期转换为unix_timestamps，我能够得到答案......在PHP中进行最终计算。如果这是错误的...请发布，我将删除答案...

SELECT
        COUNT(*) AS N,
        SUM(UNIX_TIMESTAMP(LOGENTRY.DATE)) AS Sum_X,
        SUM(UNIX_TIMESTAMP(LOGENTRY.DATE) * UNIX_TIMESTAMP(LOGENTRY.DATE)) AS Sum_X2,
        SUM(LOGENTRY.COST) AS Sum_Y,
        SUM(LOGENTRY.COST*LOGENTRY.COST) AS Sum_Y2,
        SUM(UNIX_TIMESTAMP(LOGENTRY.DATE) * LOGENTRY.COST) AS Sum_XY
    FROM LOGENTRY

MySQL的斜率（趋势）单场（最佳拟合线）

1 个答案: