Oracle SQL 中的案例表达式（Pentaho 报表设计器）

Question

我需要提取客户的年龄并将结果细分为类别（18-21、22-35、36-50 等）。

使用基于 Oracle 的 Pentaho Report Designer。

我可以计算特定年龄，但希望计数反映年龄范围，而不是单个年龄。尝试通过 case 语句完成此操作但不断出错。

最少 SQL 曝光，Pentaho/Oracle 新手，本网站新手。

--SQL 年龄和计数。

SELECT COUNT(*), "CLIENT_TABLE"."AGE"
FROM "CLIENT_TABLE"
GROUP BY "CLIENT_TABLE"."AGE"
ORDER BY "CLIENT_TABLE"."AGE"

--这是我的CASE表达式。

CASE
    WHEN "CLIENT_TABLE"."AGE" < 18 THEN 'Under 18'
    WHEN "CLIENT_TABLE"."AGE" > 18 AND <= 21 THEN '19 - 21'
    WHEN "CLIENT_TABLE"."AGE" > 21 AND <= 35 THEN '22 - 35'
END AS Age

我已将 CASE 表达式放在 SELECT 和 WHERE 子句中，但不断收到这两条错误消息... "FROM keyword not found where expects" & "Missing expression".

更新后的代码产生错误，"not a Group By Expression."

SELECT COUNT(*),
      (CASE
        WHEN "CLIENT_TABLE"."AGE" < 18 THEN 'Under 18'
        WHEN "CLIENT_TABLE"."AGE" <= 21 THEN '18 - 21'
        WHEN "CLIENT_TABLE"."AGE" <= 35 THEN '22 - 35'
        WHEN "CLIENT_TABLE"."AGE" <= 50 THEN '36 - 50'
        WHEN "CLIENT_TABLE"."AGE" <= 64 THEN '51 - 64'
        WHEN "CLIENT_TABLE"."AGE" >= 65 THEN '65+'
       END) AS AgeRange
FROM "CLIENT_TABLE"
GROUP BY (CASE
        WHEN "CLIENT_TABLE"."AGE" < 18 THEN 'Under 18'
        WHEN "CLIENT_TABLE"."AGE" <= 21 THEN '18 - 21'
        WHEN "CLIENT_TABLE"."AGE" <= 35 THEN '22 - 35'
        WHEN "CLIENT_TABLE"."AGE" <= 50 THEN '36 - 50'
        WHEN "CLIENT_TABLE"."AGE" <= 64 THEN '51 - 64'
        WHEN "CLIENT_TABLE"."AGE" >= 65 THEN '65+'
       END)
ORDER BY "CLIENT_TABLE"."AGE"

Answer 1

我想你想要：

SELECT COUNT(*),
       (CASE WHEN "CLIENT_TABLE"."AGE" < 18 THEN 'Under 18'
             WHEN "CLIENT_TABLE"."AGE" <= 21 THEN '19 - 21'
             WHEN "CLIENT_TABLE"."AGE" <= 35 THEN '22 - 35'
        END) AS Age
FROM "CLIENT_TABLE"
GROUP BY (CASE WHEN "CLIENT_TABLE"."AGE" < 18 THEN 'Under 18'
               WHEN "CLIENT_TABLE"."AGE" <= 21 THEN '19 - 21'
               WHEN "CLIENT_TABLE"."AGE" <= 35 THEN '22 - 35'
           END)
ORDER BY "CLIENT_TABLE"."AGE"

备注：

在Oracle中，需要重复GROUP BY中的表达式。
我简化了 case 表达式中条件的逻辑。条件按顺序求值。
在这样做时，18 岁的孩子也被包括在内——现在在 '19 - 21' 组中。将它们包括在某个地方似乎比将它们放在 NULL 组中更合适。

Answer 2

根据@Gordon 的回答，您的 'updated code' 得到 "ORA-00979: not a GROUP BY expression"，因为您尝试按 "CLIENT_TABLE"."AGE" 订购 - 而该列不在 GROUP BY 中条款。你可能想要：

ORDER BY MIN("CLIENT_TABLE"."AGE")

但是，这只会向您显示具有匹配数据的范围。如果您想查看所有范围，包括零计数，您可以将范围生成为内联视图或 CTE 和外部连接到您的真实数据，例如：

WITH "RANGES" ("MIN_AGE", "MAX_AGE", "LABEL") AS (
  SELECT 0, 17, 'Under 18' FROM "DUAL"
  UNION ALL
  SELECT 18, 21, '18 - 21' FROM "DUAL"
  UNION ALL
  SELECT 22, 35, '22 - 35' FROM "DUAL"
  UNION ALL
  SELECT 36, 50, '36 - 50' FROM "DUAL"
  UNION ALL
  SELECT 51, 64, '51 - 64' FROM "DUAL"
  UNION ALL
  SELECT 65, NULL, '65+' FROM "DUAL"
)
SELECT COUNT("CLIENT_TABLE"."AGE"),
       "RANGES"."LABEL" AS AgeRange
FROM "RANGES"
LEFT JOIN "CLIENT_TABLE"
ON "CLIENT_TABLE"."AGE" >= "RANGES"."MIN_AGE"
AND ("RANGES"."MAX_AGE" IS NULL OR "CLIENT_TABLE"."AGE" <= "RANGES"."MAX_AGE")
GROUP BY "RANGES"."MIN_AGE", "RANGES"."LABEL"
ORDER BY "RANGES"."MIN_AGE"

db<>fiddle

您可以根据 min/max 年龄生成标签，甚至可以只根据最小年龄生成标签（使用 lead 找到最大值），您还可以探索其他变体。

（我假设您有大写和带引号的标识符是有原因的，可能是您的报告工具的要求——我知道有些奇怪的事情确实需要它们；但从 Oracle 的角度来看，双引号不是需要，我个人觉得小写或混合代码（也许关键字大写）更容易阅读。您可能还想使用 table 别名。）

I should have an ELSE in the CASE for null values.

COUNT() 忽略空值，因此如果这些为空值，COUNT("CLIENT_TABLE"."AGE") 将 return 归零。您需要计算 table 中的任何非空列； RANGES CTE 中需要另一个虚拟行，并修改逻辑以包含未设置年龄的行；例如：

with ranges (min_age, max_age, label) as (
  select 0, 17, 'Under 18' from dual
  union all
  select 18, 21, '18 - 21' from dual
  union all
  select 22, 35, '22 - 35' from dual
  union all
  select 36, 50, '36 - 50' from dual
  union all
  select 51, 64, '51 - 64' from dual
  union all
  select 65, null, '65+' from dual
  union all
  select null, null, 'Unknown' from dual
)
select count(client_table.id), -- count any not-null column from this table
       ranges.label as agerange
from ranges
left join client_table
on ((ranges.min_age is null and client_table.age is null)
  or client_table.age >= ranges.min_age)
and (ranges.max_age is null or client_table.age <= ranges.max_age)
group by ranges.min_age, ranges.label
order by ranges.min_age

db<>fiddle

Answer 3

让我们简化一下：

SELECT AGERANGE, COUNT(1) AS CNT FROM
SELECT (CASE
        WHEN "CLIENT_TABLE"."AGE" < 18 THEN 'Under 18'
        WHEN "CLIENT_TABLE"."AGE" <= 21 THEN '18 - 21'
        WHEN "CLIENT_TABLE"."AGE" <= 35 THEN '22 - 35'
        WHEN "CLIENT_TABLE"."AGE" <= 50 THEN '36 - 50'
        WHEN "CLIENT_TABLE"."AGE" <= 64 THEN '51 - 64'
        WHEN "CLIENT_TABLE"."AGE" >= 65 THEN '65+'
       END) AS AgeRange
FROM "CLIENT_TABLE")
GROUP BY AGERANGE
ORDER BY CASE AGERANGE
WHEN 'Under 18' THEN 1 
WHEN '18 - 21' THEN 2
WHEN '22 - 35' THEN 3
WHEN '36 - 50' THEN 4
WHEN '51 - 64' THEN 5
WHEN '65+' THEN 6 
ELSE 7 END;

干杯！！

Oracle SQL 中的案例表达式（Pentaho 报表设计器）

Case Expressions in Oracle SQL (Pentaho Report Designer)

sql

oracle

pentaho-report-designer