歷史拉鍊表的應用--餘額歷史中每日平均餘額的算法

在數據倉庫中，協議餘額歷史等會以拉鍊表的形式存儲數據，以節省空間；但對拉鍊表的應用，如計算每日平均餘額要比每日全量表麻煩一些。本文針對餘額歷史表中每日平均餘額的算法作一個深入的學習。

1、  建表及插入拉鍊表測試數據

 --1、歷史拉鍊表建表 -- Create table create table EDW_T100_BAL_H ( ID VARCHAR2(8) not null, START_DATE DATE not null, BAL NUMBER(22,2), END_DATE DATE not null ); -- Add comments to the columns comment on column EDW_T100_BAL_H.ID is 'ID'; comment on column EDW_T100_BAL_H.START_DATE is '開始日期'; comment on column EDW_T100_BAL_H.BAL is '餘額'; comment on column EDW_T100_BAL_H.END_DATE is '結束日期'; -- Create/Recreate primary, unique and foreign key constraints alter table EDW_T100_BAL_H add constraint EDW_T100_BAL_H primary key (ID, START_DATE); --2、拉鍊表測試數據的插入 insert into edw_t100_bal_h (ID, START_DATE, BAL, END_DATE) values ('1', to_date('01-01-2011', 'dd-mm-yyyy'), 8.00, to_date('03-01-2011', 'dd-mm-yyyy')); insert into edw_t100_bal_h (ID, START_DATE, BAL, END_DATE) values ('1', to_date('03-01-2011', 'dd-mm-yyyy'), 10.00, to_date('08-01-2011', 'dd-mm-yyyy')); insert into edw_t100_bal_h (ID, START_DATE, BAL, END_DATE) values ('1', to_date('08-01-2011', 'dd-mm-yyyy'), 9.00, to_date('10-02-2011', 'dd-mm-yyyy')); insert into edw_t100_bal_h (ID, START_DATE, BAL, END_DATE) values ('2', to_date('01-01-2011', 'dd-mm-yyyy'), 18.00, to_date('01-02-2011', 'dd-mm-yyyy')); insert into edw_t100_bal_h (ID, START_DATE, BAL, END_DATE) values ('3', to_date('01-01-2011', 'dd-mm-yyyy'), 6.00, to_date('06-01-2011', 'dd-mm-yyyy')); commit; --3、創建臨時表(TIL=>Time interval 時間區段) -- Create table create global temporary table TMP_T100_BAL_H_TIL ( ID VARCHAR2(8) not null, START_DATE DATE not null, BAL NUMBER(22,2), END_DATE DATE not null ) on commit delete rows; --4、創建存放最終數據的目標表 -- Create table create table EDW_T100_PER_AVG_BAL_TIL ( ID VARCHAR2(8) not null, INTERVAL_START_DATE DATE not null, INTERVAL_END_DATE DATE not null, INTERVAL_DAYS NUMBER, INTERVAL_SUM_BAL NUMBER(22,2), INTERVAL_PERDAY_AVG_BAL NUMBER(22,2) ); -- Add comments to the table comment on table EDW_T100_PER_AVG_BAL_TIL is '區段每日平均餘額表'; -- Add comments to the columns comment on column EDW_T100_PER_AVG_BAL_TIL.ID is 'ID'; comment on column EDW_T100_PER_AVG_BAL_TIL.INTERVAL_START_DATE is '區段開始日期'; comment on column EDW_T100_PER_AVG_BAL_TIL.INTERVAL_END_DATE is '區段結束日期'; comment on column EDW_T100_PER_AVG_BAL_TIL.INTERVAL_DAYS is '區段天數據'; comment on column EDW_T100_PER_AVG_BAL_TIL.INTERVAL_SUM_BAL is '區段ID總餘額'; comment on column EDW_T100_PER_AVG_BAL_TIL.INTERVAL_PERDAY_AVG_BAL is '區段ID每日平均餘額'; -- Create/Recreate primary, unique and foreign key constraints alter table EDW_T100_PER_AVG_BAL_TIL add constraint P_EDW_T100_PER_AVG_BAL_TIL primary key (ID, INTERVAL_START_DATE, INTERVAL_END_DATE);

2、  實現餘額每日平均算法的代碼

 CREATE OR REPLACE PROCEDURE SP_H_TAB_PER_AVG_BAL( P_INTERVAL_START_DATE VARCHAR2, P_INTERVAL_END_DATE VARCHAR2) IS BEGIN --目標表數據恢復 DELETE EDW_T100_PER_AVG_BAL_TIL WHERE INTERVAL_START_DATE = TO_DATE(P_INTERVAL_START_DATE, 'YYYYMMDD') AND INTERVAL_END_DATE = TO_DATE(P_INTERVAL_END_DATE, 'YYYYMMDD'); --根據統計區間截取歷史拉鍊表，插入到時臨時表TMP_T100_BAL_H_TIL中 INSERT INTO TMP_T100_BAL_H_TIL (ID --ID ,START_DATE --開始日期 ,BAL --餘額 ,END_DATE --結束日期 ) SELECT A.ID ,CASE WHEN A.START_DATE <= TO_DATE(P_INTERVAL_START_DATE, 'YYYYMMDD') THEN TO_DATE(P_INTERVAL_START_DATE, 'YYYYMMDD') ELSE A.START_DATE END START_DATE --開始時間小於區間開始時間的取區間開始時間，否則取開始時間 ,A.BAL ,CASE WHEN A.END_DATE >= TO_DATE(P_INTERVAL_END_DATE, 'YYYYMMDD') THEN TO_DATE(P_INTERVAL_END_DATE, 'YYYYMMDD') + 1 ELSE A.END_DATE --結束時間大於等區間結束時間的取區間結束時間，否則取結束時間 END END_DATE FROM EDW_T100_BAL_H A WHERE (A.START_DATE < TO_DATE(P_INTERVAL_END_DATE, 'YYYYMMDD') AND A.END_DATE >= TO_DATE(P_INTERVAL_START_DATE, 'YYYYMMDD')) --過濾條件：歷史表中開始時間小於區間結束時間且歷史表中結束時間大於等於區間開始時間 ; --根據臨時表中的數據，統計區間ID的餘額總值與餘額每日平均值，插入到目標表EDW_T100_PER_AVG_BAL_TIL中 INSERT INTO EDW_T100_PER_AVG_BAL_TIL (ID --ID ,INTERVAL_START_DATE --區段開始日期 ,INTERVAL_END_DATE --區段結束日期 ,INTERVAL_DAYS --區段天數據 ,INTERVAL_SUM_BAL --區段ID總餘額 ,INTERVAL_PERDAY_AVG_BAL --區段ID每日平均餘額 ) SELECT A.ID ,TO_DATE(P_INTERVAL_START_DATE, 'YYYYMMDD') ,TO_DATE(P_INTERVAL_END_DATE, 'YYYYMMDD') ,TO_DATE(P_INTERVAL_END_DATE, 'YYYYMMDD') + 1 - TO_DATE(P_INTERVAL_START_DATE, 'YYYYMMDD') ,SUM((A.END_DATE - A.START_DATE) * A.BAL) --區間總餘額 ,SUM((A.END_DATE - A.START_DATE) * A.BAL) / (TO_DATE(P_INTERVAL_END_DATE, 'YYYYMMDD') + 1 -TO_DATE(P_INTERVAL_START_DATE, 'YYYYMMDD')) --區間總餘額/統計天數 FROM TMP_T100_BAL_H_TIL A GROUP BY A.ID; COMMIT; END; /

說明：以上的方法是根據歷史表直接算出每日平均餘額；其實，我們也可以根據歷史拉鍊表先展開爲一個每日全量表(展開方法見：http://blog.csdn.net/nsj820/archive/2010/12/16/6080701.aspx)，再根據展開後的全量表進行每日平均餘額的計算。如下：

(1)、歷史表展開爲全量表的代碼(建表語句等略)

 DECLARE CURSOR LSLL_ID IS SELECT ID, START_DATE, BAL, DECODE(END_DATE, TO_DATE('30001231', 'YYYYMMDD'), SYSDATE, --TO_DATE(P_ETLDATE, 'YYYYMMDD'), END_DATE) END_DATE FROM edw_t100_bal_h; V_ID VARCHAR2(1000); V_START_DATE DATE; V_END_DATE DATE; V_DATA_DATE DATE; V_BAL VARCHAR2(2); BEGIN BEGIN EXECUTE IMMEDIATE 'TRUNCATE TABLE EDW_T100_BAL_ALL'; OPEN LSLL_ID; LOOP FETCH LSLL_ID INTO V_ID, V_START_DATE, V_BAL, V_END_DATE; EXIT WHEN LSLL_ID%NOTFOUND; INSERT INTO EDW_T100_BAL_ALL (ID, DATA_DATE, BAL) SELECT V_ID, V_START_DATE + ROWNUM - 1 AS DATA_DATE, V_BAL FROM SYS_MATIAN_DATE WHERE V_START_DATE + ROWNUM - 1 < V_END_DATE; COMMIT; END LOOP; CLOSE LSLL_ID; END; END;

(2)、根據全量表進行每日平均餘額的計算

 SELECT ID, SUM(NVL(BAL, 0)) 餘額總值, SUM(NVL(BAL, 0)) / 391 平均每日餘額 FROM EDW_T100_BAL_ALL WHERE DATA_DATE BETWEEN TO_DATE('20100101', 'YYYYMMDD') AND TO_DATE('20110126', 'YYYYMMDD') GROUP BY ID;