Oracle分析函數、多維函數和Model函數簡要說明,主要針對BI報表統計

以下代碼均經過測試,可直接運行
Oracle分析函數、多維函數和Model函數簡要說明,主要針對BI報表統計,不一定很全面,但對BI應用場景做了少許說明

--創建一張銷售數量表,數據趨勢是遞增的
CREATE TABLE ComputerSales AS  
SELECT
 120+TRUNC(rn/12)+ROUND(DBMS_RANDOM.VALUE(1,10)) SalesNumber
  FROM
  (
    SELECT level,ROWNUM rn
      FROM DUAL
   CONNECT BY ROWNUM<=120
  );

--下面用於比較NULL值和非NULL值的統計,可以看出NULL值情況下的COUNT是存在問題的,所以建議數據庫系統中最好不要使用NULL值列
SELECT
  COUNT(*),
  COUNT(a.SalesNumber),
  COUNT(DISTINCT a.SalesNumber),
  SUM(a.SalesNumber),
  AVG(a.SalesNumber),
  MAX(a.SalesNumber),
  MIN(a.SalesNumber)
  FROM ComputerSales A;
DELETE FROM ComputerSales WHERE SalesNumber IS NULL;
COMMIT;
INSERT INTO ComputerSales VALUES(NULL);
COMMIT;
INSERT INTO ComputerSales VALUES(NULL);
COMMIT;
SELECT
  COUNT(*),
  COUNT(a.SalesNumber),
  COUNT(DISTINCT a.SalesNumber),
  SUM(a.SalesNumber),
  AVG(a.SalesNumber),
  MAX(a.SalesNumber),
  MIN(a.SalesNumber)
  FROM ComputerSales A;
SELECT trunc(dbms_random.value(1,101)), 


DELETE FROM ComputerSales WHERE SalesNumber IS NULL;
COMMIT;
--創建增加了日期字段的表
CREATE TABLE ComputerSalesBAK AS  
SELECT SalesNumber,TRUNC(SYSDATE)+MOD(A.DateSEQ-1,10) SalesDate
  FROM (SELECT SalesNumber,ROW_NUMBER() OVER(ORDER BY ROWID) DateSEQ FROM ComputerSales) A;
DROP TABLE ComputerSales;
RENAME ComputerSalesBAK TO ComputerSales;

--下面是兩種創建方式,構招Area列和日期列
CREATE TABLE ComputerSalesBAK AS  
SELECT SalesNumber,TRUNC(SYSDATE)+MOD(A.DateSEQ-1,24) SalesDate,
       CASE WHEN TRUNC((DateSEQ-1)/24)=1 THEN '華南地區'
            WHEN TRUNC((DateSEQ-1)/24)=2 THEN '華北地區'
            WHEN TRUNC((DateSEQ-1)/24)=3 THEN '東北地區'
            WHEN TRUNC((DateSEQ-1)/24)=4 THEN '華東地區'
            ELSE '其他地區'
       END
  FROM (SELECT SalesNumber,ROW_NUMBER() OVER(ORDER BY ROWID) DateSEQ FROM ComputerSales) A;
DROP TABLE ComputerSales;
RENAME ComputerSalesBAK TO ComputerSales;

--該例可構造SalesDate和Area的重複數據
CREATE TABLE ComputerSalesBAK AS
SELECT SalesNumber,
       TRUNC(SYSDATE)+MOD(A.DateSEQ-1,10) SalesDate,
       CASE WHEN AreaSEQ=1 THEN '華南地區'
            WHEN AreaSEQ=2 THEN '華北地區'
            WHEN AreaSEQ=3 THEN '東北地區'
            WHEN AreaSEQ=4 THEN '華東地區'
            ELSE '其他地區'
       END
  FROM (SELECT SalesNumber,ROW_NUMBER() OVER(ORDER BY ROWID) DateSEQ,ROUND(dbms_random.VALUE(1,5)) AreaSEQ FROM ComputerSales) A;
DROP TABLE ComputerSales;
RENAME ComputerSalesBAK TO ComputerSales;
 

--移動平均值,累計求和,當前窗口平均值,當前窗口求和,以及窗口函數和排序函數的作用域
SELECT
  Area,SalesDate,SalesNumber,
  MIN(SalesNumber) OVER (PARTITION BY Area order by SalesDate) AS min_Area_SalesDate,
  MAX(SalesNumber) OVER (PARTITION BY Area order by SalesDate) AS max_Area_SalesDate,
  AVG(SalesNumber) OVER (PARTITION BY Area order by SalesDate) AS avg_Area_SalesDate,  
  SUM(SalesNumber) OVER (PARTITION BY Area order by SalesDate) AS sum_Area_SalesDate,  
  COUNT(*) OVER (PARTITION BY Area ORDER BY SalesDate) AS count_Area,
  MIN(SalesNumber) OVER (PARTITION BY Area) AS min_Area,
  MAX(SalesNumber) OVER (PARTITION BY Area) AS max_Area,
  AVG(SalesNumber) OVER (PARTITION BY Area) AS avg_Area,  
  SUM(SalesNumber) OVER (PARTITION BY Area) AS sum_Area,  
  COUNT(*) OVER (PARTITION BY Area) AS count_Area 
FROM ComputerSales

--觀察Rank、Dense_Rank,Row_number,Count的區別
--Rank跳號,Dense_Rank不跳號,Row_number唯一,Count按統計數計也跳號
--如果PARTITION BY和order by 的字段是唯一的話,則這四個函數沒什麼區別
SELECT
  Area,SalesDate,SalesNumber,
  RANK() OVER (PARTITION BY Area order by SalesNumber) AS Rank_Area_SalesNumber,
  DENSE_RANK() OVER (PARTITION BY Area order by SalesNumber) AS DenseRank_Area_SalesNumber,
  ROW_NUMBER() OVER (PARTITION BY Area order by SalesNumber) AS Rownumber_Area_SalesNumber,
  COUNT(*) OVER (PARTITION BY Area order by SalesNumber) AS CountAll_Area_SalesNumber,
  COUNT(SalesNumber) OVER (PARTITION BY Area order by SalesNumber) AS Count_Area_SalesNumber
FROM ComputerSales

--觀察Lag和Lead的異同,以及Lag參數之間的異同
--缺省情況下Lag取前一行的值,Lead取後一行的值
--Lag、lead的第一個參數決定了取行的位置,第二個參數爲取不到值時的缺省值
SELECT
  Area,SalesDate,SalesNumber,
  LAG(SalesNumber) OVER (PARTITION BY Area order by SalesDate) AS Lag_Area_SalesNumber, 
  LEAD(SalesNumber) OVER (PARTITION BY Area order by SalesDate) AS Lead_Area_SalesNumber,   
  LAG(SalesNumber,1) OVER (PARTITION BY Area order by SalesDate) AS Lag1_Area_SalesNumber,
  LAG(SalesNumber,2) OVER (PARTITION BY Area order by SalesDate) AS Lag2_Area_SalesNumber,
  LEAD(SalesNumber,1) OVER (PARTITION BY Area order by SalesDate) AS Lead1_Area_SalesNumber,
  LEAD(SalesNumber,2) OVER (PARTITION BY Area order by SalesDate) AS Lead2_Area_SalesNumber,
  LAG(SalesNumber,1,0) OVER (PARTITION BY Area order by SalesDate) AS Lag10_Area_SalesNumber,
  LAG(SalesNumber,2,1) OVER (PARTITION BY Area order by SalesDate) AS Lag21_Area_SalesNumber,
  LEAD(SalesNumber,1,0) OVER (PARTITION BY Area order by SalesDate) AS Lead10_Area_SalesNumber,
  LEAD(SalesNumber,2,1) OVER (PARTITION BY Area order by SalesDate) AS Lead21_Area_SalesNumber 
FROM ComputerSales

--觀察First_Value和Last_Value的不同
--如果取同一個同組中最大值最小值對應的某列,使用FIRST_VALUE,按照升降序排列即可
--LAST_VALUE有些像兩次分組所求的最後一行
SELECT
  Area,SalesDate,SalesNumber,
  FIRST_VALUE(SalesDate) OVER (PARTITION BY Area order by SalesNumber) AS FirstValue_Area, 
  FIRST_VALUE(SalesDate) OVER (PARTITION BY Area order by SalesNumber DESC) AS FirstValue_Area_Desc,   
  LAST_VALUE(SalesDate) OVER (PARTITION BY Area order by SalesNumber) AS LastValue_Area,
  LAST_VALUE(SalesDate) OVER (PARTITION BY Area order by SalesNumber DESC) AS LastValue_Area_Desc
FROM ComputerSales

--與上面不同的是,KEEP需要和DENSE_RANK FIRST |DENSE_RANK LAST配合使用,且取的是相同Area中按SalesNumber排序所獲得最大或最小的值,而上面只是取第一行或最後一行
SELECT Area,SalesDate,SalesNumber,
  DENSE_RANK() OVER(PARTITION BY Area ORDER BY SalesNumber) DENSE_RANK,
  MIN(SalesDate) KEEP (DENSE_RANK FIRST ORDER BY SalesNumber) OVER(PARTITION BY Area) min_first,
  MIN(SalesDate) KEEP (DENSE_RANK LAST ORDER BY SalesNumber) OVER(PARTITION BY Area) min_last,
  MAX(SalesDate) KEEP (DENSE_RANK FIRST ORDER BY SalesNumber) OVER(PARTITION BY Area) max_first,
  MAX(SalesDate) KEEP (DENSE_RANK LAST ORDER BY SalesNumber) OVER(PARTITION BY Area) max_last
FROM ComputerSales

--CUME_DIST和PERCENT_RANK差不多,都是累計計算比例,只不過計算基準不同,CUME_DIST更符合一般的做法
--NTILE把數據平分爲若干份,更適合用來計算四分位上的值
--RATIO_TO_REPORT,則是求當前值在分區中的比例,且不能與ORDER BY 合起來使用
--PERCENTILE_DISC和PERCENTILE_CONT,則是給定的比例參數所對應的值,一般使用PERCENTILE_DISC即可
SELECT Area,SalesDate,SalesNumber,
  ROUND(CUME_DIST() OVER(PARTITION BY Area ORDER BY SalesNumber),2) cume_dist,
  ROUND(PERCENT_RANK() OVER(PARTITION BY Area ORDER BY SalesNumber),2) PERCENT_RANK,
  ROUND(RATIO_TO_REPORT(SalesNumber) OVER(PARTITION BY Area),2) RATIO_TO_REPORT,
  NTILE(4) OVER(PARTITION BY Area ORDER BY SalesNumber) NTILE,
  PERCENTILE_DISC(0.7) WITHIN GROUP (ORDER BY SalesNumber) OVER(PARTITION BY Area) PERCENTILE_DISC,
  PERCENTILE_CONT(0.7) WITHIN GROUP (ORDER BY SalesNumber) OVER(PARTITION BY Area) PERCENTILE_CONT
FROM ComputerSales

--增加了一列叫銷售額,可以進行相關數理統計
CREATE TABLE ComputerSalesBAK AS  
SELECT SalesNumber,
       ROUND(SalesNumber*10+5*DBMS_RANDOM.VALUE(1,10)) SalesValue,
       TRUNC(SYSDATE)+MOD(A.DateSEQ-1,24) SalesDate,
       CASE WHEN TRUNC((DateSEQ-1)/24)=1 THEN '華南地區'
            WHEN TRUNC((DateSEQ-1)/24)=2 THEN '華北地區'
            WHEN TRUNC((DateSEQ-1)/24)=3 THEN '東北地區'
            WHEN TRUNC((DateSEQ-1)/24)=4 THEN '華東地區'
            ELSE '其他地區'
       END Area
  FROM (SELECT SalesNumber,ROW_NUMBER() OVER(ORDER BY ROWID) DateSEQ FROM ComputerSales) A;
DROP TABLE ComputerSales;
RENAME ComputerSalesBAK TO ComputerSales;
SELECT * FROM ComputerSales;

--其他統計,對數理分析有研究的同學可以嘗試一下其經濟學含義
SELECT Area,SalesDate,SalesValue,SalesNumber,
  REGR_SLOPE(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "斜率",
  REGR_INTERCEPT(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "截距",
  REGR_R2(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "迴歸線決定係數",
  REGR_AVGX(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "迴歸線自變量平均值",
  REGR_AVGY(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "迴歸線應變量平均值", 
  VAR_POP(SalesValue) OVER(PARTITION BY Area ORDER BY SalesDate) "VAR_POP_應變量", 
  VAR_POP(SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "VAR_POP_自變量", 
  COVAR_POP(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "COVAR_POP",       
  REGR_SXX(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "REGR_SXX",  --REGR_COUNT(expr1, expr2) * VAR_POP(expr2) 
  REGR_SYY(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "REGR_SXY",  --REGR_COUNT(expr1, expr2) * VAR_POP(expr1)
  REGR_SXY(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "REGR_SXY",  --REGR_COUNT(expr1, expr2) * COVAR_POP(expr1, expr2)  
  REGR_COUNT(SalesValue,SalesNumber) OVER(PARTITION BY Area ORDER BY SalesDate) "REGR_COUNT"
FROM ComputerSales

--關於按日期進行環比的問題
--同比則有麻煩,因爲日期天數是不固定的
--從ComputerSales隨機刪除幾行再測
SELECT AREA,SALESDATE,SALESNUMBER,
  LAG(SalesNumber) OVER (PARTITION BY Area order by SalesDate) AS Lag_error,  --如遇斷號,會導致數據不準
  SUM(SalesNumber) OVER (PARTITION BY AREA ORDER BY SALESDATE RANGE BETWEEN 1 PRECEDING AND 1 PRECEDING) yesterday, --昨天的值 
  SUM(SalesNumber) OVER (PARTITION BY AREA ORDER BY SALESDATE RANGE BETWEEN 6 PRECEDING AND 6 PRECEDING) lastweek, --上週數據 
  SUM(SalesNumber) OVER (PARTITION BY AREA ORDER BY SALESDATE RANGE BETWEEN 6 PRECEDING AND 0 PRECEDING) last7_accu, --前7天累計,包括當天
  SUM(SalesNumber) OVER (PARTITION BY AREA ORDER BY SALESDATE RANGE BETWEEN 29 PRECEDING AND 0 PRECEDING) last30_accu--前30天累計,包括當天
  FROM ComputerSales
 
--再度增加一個product產品列,以方便進行CUBE函數演示
CREATE TABLE ComputerSalesBAK AS  
SELECT SalesNumber,
       ROUND(SalesNumber*10+5*DBMS_RANDOM.VALUE(1,10)) SalesValue,
       TRUNC(SYSDATE)+MOD(A.DateSEQ-1,24) SalesDate,
       CASE WHEN TRUNC((DateSEQ-1)/24)=1 THEN '華南地區'
            WHEN TRUNC((DateSEQ-1)/24)=2 THEN '華北地區'
            WHEN TRUNC((DateSEQ-1)/24)=3 THEN '東北地區'
            WHEN TRUNC((DateSEQ-1)/24)=4 THEN '華東地區'
            ELSE '其他地區'
       END Area,
       CASE WHEN ROUND(DBMS_RANDOM.VALUE(1,3))=1 THEN '產品A'
            WHEN ROUND(DBMS_RANDOM.VALUE(1,3))=2 THEN '產品B'
            ELSE '產品C'
       END Product      
  FROM (SELECT SalesNumber,ROW_NUMBER() OVER(ORDER BY ROWID) DateSEQ FROM ComputerSales) A;
DROP TABLE ComputerSales;
RENAME ComputerSalesBAK TO ComputerSales;
SELECT * FROM ComputerSales;

--傳統的group by語法
SELECT Product,Area,SalesDate,SUM(SalesNumber),SUM(SalesValue)
  FROM ComputerSales
 GROUP BY Product,Area,SalesDate
 ORDER BY Product,Area,SalesDate
 
--ROLLUP (group的字段順序)
--會自動按Group字段分層統計,與日常報表較爲相似
SELECT Product,Area,SalesDate,SUM(SalesNumber),SUM(SalesValue)
  FROM ComputerSales
 GROUP BY ROLLUP(Product,Area,SalesDate)
 ORDER BY Product,Area,SalesDate --加不加均可,已經自動按分組字段排序
 
--等價於
SELECT * FROM
(
SELECT Product,Area,SalesDate,SUM(SalesNumber) SalesNumber,SUM(SalesValue) SalesValue --最大級分組
  FROM ComputerSales
 GROUP BY Product,Area,SalesDate
 UNION ALL
SELECT Product,Area,NULL,SUM(SalesNumber),SUM(SalesValue) --按產品、地區分組
  FROM ComputerSales
 GROUP BY Product,Area,NULL
 UNION ALL
SELECT Product,NULL,NULL,SUM(SalesNumber),SUM(SalesValue) --按產品分組
  FROM ComputerSales
 GROUP BY Product,NULL,NULL
 UNION ALL 
SELECT NULL,NULL,NULL,SUM(SalesNumber),SUM(SalesValue)   --統計總和
  FROM ComputerSales
 GROUP BY NULL,NULL,NULL
) ORDER BY 1,2,3                                         --最後再排序
 
 
--CUBE (group的字段順序),與OLAP比較相似,求得所有維度的交匯點
--會自動按Group字段排列組合進行統計
SELECT Product,Area,SalesDate,SUM(SalesNumber),SUM(SalesValue)
  FROM ComputerSales
 GROUP BY CUBE(Product,Area,SalesDate)
 ORDER BY Product,Area,SalesDate --加不加均可,已經自動按分組字段排序
--兩則的區別
--即ROLLUP 爲C(3,1)即多了3層
--按照Product,Area,SalesDate;Product,Area;Product;ALL的順序進行了統計
--CUBE的統計層級則爲2的N次方,即全部的有序組合
--按照Product,Area,SalesDate;Product,Area;Product,SalesDate;Product;Area,SalesDate;Area;SalesDate;ALL的順序進行了統計
--與ROLLUP的等價表達式,相當於ROLLUP的排列組合
SELECT * FROM
(
SELECT Product,Area,SalesDate,SUM(SalesNumber),SUM(SalesValue) --先按Product,Area,SalesDate求ROLLUP
  FROM ComputerSales
 GROUP BY ROLLUP(Product,Area,SalesDate)
UNION
SELECT Product,NULL,SalesDate,SUM(SalesNumber),SUM(SalesValue) --再按Product,SalesDate求ROLLUP
  FROM ComputerSales
 GROUP BY ROLLUP(Product,NULL,SalesDate)
UNION
SELECT NULL,Area,SalesDate,SUM(SalesNumber),SUM(SalesValue) --再按Area,SalesDate求ROLLUP
  FROM ComputerSales
 GROUP BY ROLLUP(NULL,Area,SalesDate)
UNION
SELECT NULL,NULL,SalesDate,SUM(SalesNumber),SUM(SalesValue) --最後按SalesDate求ROLLUP
  FROM ComputerSales
 GROUP BY ROLLUP(NULL,NULL,SalesDate)
 )
 ORDER BY 1,2,3

--GROUPING SETS等同於按三列單獨求統計,一般不常用
SELECT Product,Area,SalesDate,SUM(SalesNumber),SUM(SalesValue)
  FROM ComputerSales
 GROUP BY GROUPING SETS(Product,Area,SalesDate)
 ORDER BY Product,Area,SalesDate ;--加不加均可,已經自動按分組字段排序
--等價於
SELECT * FROM
(
SELECT Product,NULL Area,NULL SalesDate,SUM(SalesNumber),SUM(SalesValue) --按產品分組
  FROM ComputerSales
 GROUP BY Product,NULL,NULL
 UNION ALL
SELECT NULL,Area,NULL,SUM(SalesNumber),SUM(SalesValue) --按地區分組
  FROM ComputerSales
 GROUP BY NULL,Area,NULL
 UNION ALL
SELECT NULL,NULL,SalesDate,SUM(SalesNumber) SalesNumber,SUM(SalesValue) SalesValue --按日期分組
  FROM ComputerSales
 GROUP BY NULL,NULL,SalesDate
) ORDER BY 1,2,3    

--GROUPING函數只接受一個參數,參數爲數據表的一列。如果該列爲空返回1,否則返回0。
--並且它僅能與 GROUP BY,ROLLUP,CUBE,GROUPING SETS 一起使用。
--稍微運行一下,就發現該函數只是爲了做BI報表使用的,把統計行變爲1,將來用作字符串替代
SELECT GROUPING(Product), Product,GROUPING(Area),Area,GROUPING(SalesDate),SalesDate,SUM(SalesNumber),SUM(SalesValue)
  FROM ComputerSales
 GROUP BY ROLLUP(Product,Area,SalesDate)
 ORDER BY Product,Area,SalesDate ;
--BI標準報表格式
SELECT
  DECODE(ProductFlag,1,'產品彙總',Product),
  DECODE(AreaFlag,1,'地區彙總',Area),
  DECODE(SalesDateFlag,1,'日期彙總',TO_CHAR(SalesDate,'YYYY-MM-DD')),
  SalesNumber,SalesValue
  FROM
(
SELECT
  GROUPING(Product) ProductFlag, Product,
  GROUPING(Area) AreaFlag,Area,
  GROUPING(SalesDate) SalesDateFlag,SalesDate,
  SUM(SalesNumber) SalesNumber,SUM(SalesValue) SalesValue
  FROM ComputerSales
 GROUP BY ROLLUP(Product,Area,SalesDate)
 ORDER BY Product,Area,SalesDate
)

--GROUPING_ID其實和GROUPING原理差不多,GROUPING參數爲單值,且只返回1,1
--GROUPING_ID,則返回按2的指數進行累計得到空值區域的值
SELECT Product,Area,SalesDate,
       GROUPING_ID(Product,Area,SalesDate) GROUPING421,
       GROUPING_ID(Product,Area) GROUPPING21,
       GROUPING_ID(Product) GROUPING1,
       SUM(SalesNumber),
       SUM(SalesValue)
  FROM ComputerSales
 GROUP BY ROLLUP(Product,Area,SalesDate)
 ORDER BY Product,Area,SalesDate ;--加不加均可,已經自動按分組字段排序
 
--GROUP_ID函數可以區分重複分組結果,第1 次出現爲0,以後每次出現增1。
--GROUP_ID單獨答應在SELECT 中出現意義不大,常在HAVING 中使用達到過濾重複統計的目的。
SELECT Product,Area,SalesDate,GROUP_ID(),
       SUM(SalesNumber),SUM(SalesValue)
  FROM ComputerSales
 GROUP BY CUBE(Product,Area),CUBE(Product,SalesDate)
HAVING GROUP_ID()=0
 ORDER BY 1,2,3
--例如該例子中分別按Product,Area和Product,SalesDate會導致產品地區、產品時間的重複計算,導致報表的不清晰
--我們用HAVING GROUP_ID()=0把重複計算的行去掉就OK了
--一般情況下不建議報表程序過度分組,否則到最後連自己都搞糊塗了
--GROUP BY,ROLLUP,CUBE能組合使用,但SELECT中的分組字段必須出現在GROUP BY的相關欄位

--MODEL:MODEL語句的關鍵字,必須。
--DIMENSION BY:DIMENSION維度的意思,可以理解爲數組的索引,必須。
--MEASURES:指定作爲數組的列
--RULES:對數組進行各種操作的描述。
--暫時還沒搞明白如何應用,只是簡單實現了一個求上月、前30天、前7天,前1天的例子
SELECT AREA,PRODUCT,SALESDATE,SALESNUMBER,
       AVG30DAY,AVG1MONTH, --最近30天的平均值,最近一個月的平均值
       ACCU30DAY,ACCU1MONTH, --最近30天的累加值,最近一個月的累加值
       SALESNUMBER1DAY,SALESNUMBER7DAY, --昨天的銷售額,一週前的銷售額
       SALESNUMBER30DAY,SALESNUMBER1MONTH  --30天的銷售額,上月同天的銷售額
  FROM ComputerSales
 MODEL DIMENSION BY (AREA,PRODUCT,SALESDATE)
 MEASURES (SALESNUMBER,0 AVG30DAY,0 AVG1MONTH,0 ACCU30DAY,0 ACCU1MONTH,0 SALESNUMBER1DAY,0 SALESNUMBER7DAY,0 SALESNUMBER30DAY,0 SALESNUMBER1MONTH)
 RULES UPDATE
 (AVG30DAY[ANY,ANY,ANY]=AVG(SALESNUMBER)[CV(),CV(),SALESDATE BETWEEN CV(SALESDATE)-29 AND CV(SALESDATE)],
  AVG1MONTH[ANY,ANY,ANY]=AVG(SALESNUMBER)[CV(),CV(),SALESDATE BETWEEN ADD_MONTHS(CV(SALESDATE),-1) AND CV(SALESDATE)],
  ACCU30DAY[ANY,ANY,ANY]=SUM(SALESNUMBER)[CV(),CV(),SALESDATE BETWEEN CV(SALESDATE)-30 AND CV(SALESDATE)],
  ACCU1MONTH[ANY,ANY,ANY]=SUM(SALESNUMBER)[CV(),CV(),SALESDATE BETWEEN ADD_MONTHS(CV(SALESDATE),-1) AND CV(SALESDATE)],
  SALESNUMBER1DAY[ANY,ANY,ANY]=MAX(SALESNUMBER)[CV(),CV(),SALESDATE BETWEEN CV(SALESDATE)-1 AND CV(SALESDATE)-1],
  SALESNUMBER7DAY[ANY,ANY,ANY]=MAX(SALESNUMBER)[CV(),CV(),SALESDATE BETWEEN CV(SALESDATE)-7 AND CV(SALESDATE)-7],
  SALESNUMBER30DAY[ANY,ANY,ANY]=MAX(SALESNUMBER)[CV(),CV(),SALESDATE BETWEEN CV(SALESDATE)-30 AND CV(SALESDATE)-30],
  SALESNUMBER1MONTH[ANY,ANY,ANY]=MAX(SALESNUMBER)[CV(),CV(),SALESDATE BETWEEN CV(SALESDATE)-30 AND CV(SALESDATE)-30] 
  )
ORDER BY 1,2,3


關於按年月環比統計中可能出現的問題

CREATE TABLE TEST (SALESMONTH VARCHAR(6),SALESNUMBER INT) ;
INSERT INTO TEST VALUES('201002',2);
INSERT INTO TEST VALUES('201004',4);
INSERT INTO TEST VALUES('201007',7);
INSERT INTO TEST VALUES('201008',8);
INSERT INTO TEST VALUES('201010',10);

SELECT SALESMONTH,SALESNUMBER,
LAG(SalesNumber) OVER(order by SalesMONTH) AS Lag10_Area_SalesNumber,
--如遇斷號,會導致數據不準
SUM(SalesNumber) OVER(ORDER BY TO_DATE(SalesMONTH||'01','YYYYMMDD') RANGE BETWEEN 1 PRECEDING AND 1 PRECEDING)
FROM TEST

遇到一個問題,假如BI報表中的月份是字符串,而碰巧斷月了,如何準確求得上個月的數據,理應爲空
如果是天的話可以想辦法規避掉,如果是字符串月沒想好怎麼處理

newkid給了算法
SELECT SALESMONTH,SALESNUMBER, 
  MAX(SalesNumber) OVER(order by TO_DATE(SalesMONTH,'YYYYMM') RANGE BETWEEN 31 PRECEDING AND 1 PRECEDING )
FROM TEST;
但我覺得結果很正確,但是不保險,而且有點迷糊
是把當前的月份轉換成當月的第一天,並且向前推31天到前1天
假如當前月是2月,向前推31天應該到去年12月份了,求的 MAX(SalesNumber) 未必有效
可實際結果是正確的,奇怪

關於Model的用法,實在讀不下去
http://download.oracle.com/docs/cd/B19306_01/server.102/b14223/sqlmodel.htm
發表評論
所有評論
還沒有人評論,想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章