将行转换为列，将列转换为行以在 BigQuery 中执行计算

Question

所以我在 google-bigquery 中有这张表，如下所示

初始表

month | gender | sales | quantity | Jan-2022 | male. | 2000. | 50. | Jan-2022 | female. | 3000. | 100. |

这是我想要的桌子

month |metrics. |overall | male |composition (%)|female|composition (%)| Jan-22|sales. | 5000. | 2000.| 40% |3000 | 60% | Jan-22|quantity. | 150. | 50.| 33% | 100 | 66% |

那么使用什么查询来获得所需的结果呢？

Answer 1

您基本上想将列转换为行并将行转换为列。 为此，您可以使用UNPIVOT和PIVOT运算符。

这为您提供了所需的 output：

 WITH your_table AS ( SELECT 'Jan-2022' AS month, 'male' AS gender, 2000 AS sales, 50 AS quantity UNION ALL SELECT 'Jan-2022', 'female', 3000, 100 ) SELECT month, metrics, (male + female) AS overall, male, male / (male + female) AS male_composition, female, female / (male + female) AS female_composition FROM ( SELECT * FROM your_table UNPIVOT(vals FOR metrics IN (sales, quantity)) ) PIVOT(SUM(vals) FOR gender IN ('male', 'female'))

Answer 2

考虑以下方法

select * from ( select *, sum(col) over month_metric overall, round(100 * (sum(col) over month_metric_gender) / (sum(col) over month_metric), 0) composition from your_table unpivot (col FOR metrics IN (sales, quantity)) window month_metric as (partition by month, metrics), month_metric_gender as (partition by month, metrics, gender)) pivot (sum(col) gender, any_value(composition) composition for gender in ('male', 'female'))

如果应用于您问题中的示例数据 - output 是

我之所以推荐上述方法而不是以前的方法（您得到的第一个答案）是因为它更通用，并且不依赖于显式调用诸如male 、 femail之类的列（当然，在 pivot 语句中除外）