按计数排序未正确排序 - SQL (MS Access 2007)

Posted 2023-03-29

【中文标题】按计数排序未正确排序 - SQL (MS Access 2007)

【英文标题】：Order by count not sorting correctly - SQL (MS Access 2007)

【发布时间】：2012-10-15 03:18:05

【问题描述】：

概述： 比较不同账户的农产品代理日常交易，找出代理和公司持股之间的差异，仅显示新产品；当前交易中的代码，但不是以前的交易。

目标： 按频率最高的代码（即苹果）对交易进行排序。

当前表：

--------------------------------------------------------------------------------------------------------------
|        DATE | ACCT_CODE | TYPE | TYPE_DESCRIPTION |    CODE | DIFFERENCE | AGENT_HOLDING | COMPANY_HOLDING |
--------------------------------------------------------------------------------------------------------------
| 02 Aug 2012 | acc107101 | prod |          produce |   apple |      -1500 |             0 |            1500 |
| 02 Aug 2012 | acc107101 | prod |          produce |   apple |       1500 |          2000 |             500 |
| 02 Aug 2012 | acc107101 | prod |          produce |   apple |          0 |             0 |               0 |
| 02 Aug 2012 | acc107101 | prod |          produce |   apple |          0 |         49797 |           49797 |
| 02 Aug 2012 | acc170000 | prod |          produce |  grapes |       2500 |          2500 |               0 |
| 02 Aug 2012 | acc170000 | prod |          produce |  grapes |      -1500 |             0 |            1500 |
| 02 Aug 2012 | acc008221 | caus |             cash |    cash |          0 |             0 |               0 |
| 02 Aug 2012 | acc008221 | caus |             cash | carrots |          0 |         43273 |           43274 |
| 01 Aug 2012 | acc008221 | caus |             cash |    cash |          0 |             0 |               0 |
| 01 Aug 2012 | acc008221 | caus |             cash | carrots |          0 |         43273 |           43274 |
| 02 Aug 2012 | acc179185 | prod |          produce |   mango |        125 |          2775 |            2650 |
| 02 Aug 2012 | acc179185 | prod |          produce |   mango |     -57646 |        453639 |          511286 |
| 02 Aug 2012 | acc636903 | prod |          produce |   mango |          0 |        481081 |          481081 |
| 02 Aug 2012 | acc001049 | prod |          produce |   mango |       2500 |          2500 |               0 |
| 02 Aug 2012 | acc001189 | prod |          produce |   apple |      -1091 |             0 |            1091 |
--------------------------------------------------------------------------------------------------------------

代码：

SELECT 
    Count(CurrentTrade.Code) AS CountOfCode, 
    CurrentTrade.date, 
    CurrentTrade.Acct_Code, 
    CurrentTrade.Type, 
    CurrentTrade.Code, 
    CurrentTrade.Agent_Holding, 
    CurrentTrade.Company_Holding, 
    CurrentTrade.Difference
FROM CurrentTrade 
LEFT JOIN PrevTrade 
    ON CurrentTrade.Company_Holding=PrevTrade.Company_Holding
WHERE (((PrevTrade.Company_Holding) Is Null) AND ((CurrentTrade.Code)<>'cash'))
GROUP BY 
    CurrentTrade.Code,CurrentTrade.date, 
    CurrentTrade.Acct_Code, 
    CurrentTrade.Type, 
    CurrentTrade.Agent_Holding, 
    CurrentTrade.Company_Holding, 
    CurrentTrade.Difference
ORDER BY count(CurrentTrade.Code) ASC;

当前结果：

---------------------------------------------
| COUNTOFCODE |        DATE | TYPE |   CODE |
---------------------------------------------
|           1 | 02 Aug 2012 | prod | grapes |
|           1 | 02 Aug 2012 | prod | grapes |
|           1 | 02 Aug 2012 | prod |  apple |
|           1 | 02 Aug 2012 | prod |  apple |
|           1 | 02 Aug 2012 | prod |  apple |
|           1 | 02 Aug 2012 | prod |  apple |
|           1 | 02 Aug 2012 | prod |  apple |
|           1 | 02 Aug 2012 | prod |  mango |
|           1 | 02 Aug 2012 | prod |  mango |
|           1 | 02 Aug 2012 | prod |  mango |
|           1 | 02 Aug 2012 | prod |  mango |
---------------------------------------------

问题： 如何修复以使用正确的 ORDER BY 频率（苹果、芒果、葡萄）？

【参考方案1】：

我认为问题在于您的分组太细了。

在这种情况下，从 GROUP BY 子句中排除三个 Amount 列（AgentHolding、CompanyHolding 和 Difference）并在输出中聚合它们（sum）可能是有意义的。

以下是此更改将产生的结果的示例：http://www.sqlfiddle.com/#!3/fc4d4/7

注意：我省略了 left join 和 PrevTrade 表，因为我不知道它的结构和数据，但希望这能说明意图。

【讨论】：

太棒了..谢谢！这更清洁并且有效。我添加了左连接，因为查询会即时比较 2 个表/2 个日期。