如何

Question

我有以下數據

sk_patient casenumber src  status pen_dt 
-1   2345  POPDB CLOSED 2012-01-01 
-1   235  POPDB CLOSED 2012-01-01 
-1   245  JCARE CLOSED 2012-01-02 
1244   2   POPDB CLOSED 2012-01-01 
273   2345  POPDB CLOSED 2012-03-04 
1244   23   JCARE CLOSED 2012-05-02 

需要被轉換爲以下不同src列值的基礎上，SRC柱的不同的價值轉移到不同的列。

sk_patient casenumber1 src1 status1 pen_dt1  casenumber2 src2 status2 pen_dt2 
    -1  2345  POPDB CLOSED  2012-01-01 245  JCARE CLOSED 2012-01-02 
    -1  235   POPDB CLOSED  2012-01-01 NULL  NULL NULL  NULL 
    1244  2   POPDB CLOSED  2012-01-01 23  JCARE CLOSED 2012-05-02 
    273  2345  POPDB CLOSED  2012-03-04 NULL  NULL NULL  NULL 

例如，

爲-1我們有3行，但他們兩個人有相同的src，因此我們將它們，因爲它是，但3行有不同的值。因此，需要將其移至src2列以及相關的列值。

將會有最小1和最大3個不同的src可以關聯任何特定的sk_patient。

===更新===

直到現在，我能夠做到以下

seq sk_patient casenumber src  status pen_dt 
2 -1   2345  POPDB CLOSED 2012-01-01 
2 -1   235  POPDB CLOSED 2012-01-01 
1 -1   245  JCARE CLOSED 2012-01-02 
2 1244   2   POPDB CLOSED 2012-01-01 
1 273   2345  POPDB CLOSED 2012-03-04 
1 1244   23   JCARE CLOSED 2012-05-02 

我用下面的查詢

select dense_rank() 
     over(partition by sk_patient order by sk_patient, src) as seq, 
     * from patient 

誰能幫我像上面的例子那樣移動值？

Answer 1

SELECT tp1.sk_patient, 
     tp1.casenumber casenumber1, 
     tp1.src src1, 
     tp1.status status1, 
     tp1.pen_dt pen_dt1, 
     tp2.casenumber2, 
     tp2.src2, 
     tp2.status2, 
     tp2.pen_dt2 
    FROM t_patient tp1 
    LEFT OUTER JOIN (SELECT sk_patient, 
          MAX(CASE WHEN rnum = 1 THEN casenumber END) casenumber1, 
          MAX(CASE WHEN rnum = 1 THEN src END) src1, 
          MAX(CASE WHEN rnum = 1 THEN status END) status1, 
          MAX(CASE WHEN rnum = 1 THEN pen_dt END) pen_dt1, 
          MAX(CASE WHEN rnum = 2 THEN casenumber END) casenumber2, 
          MAX(CASE WHEN rnum = 2 THEN src END) src2, 
          MAX(CASE WHEN rnum = 2 THEN status END) status2, 
          MAX(CASE WHEN rnum = 2 THEN pen_dt END) pen_dt2 
        FROM (SELECT *, 
            DENSE_RANK() OVER (PARTITION BY sk_patient ORDER BY src DESC) rnum 
          FROM t_patient 
         ) t 
        GROUP BY sk_patient 
       ) tp2 
    ON tp1.sk_patient = tp2.sk_patient 
    AND tp1.casenumber = tp2.casenumber1 
    AND tp1.src = tp2.src1 
WHERE tp1.src != 'JCARE' 

結果

sk_patient casenumber1 src1 status1 pen_dt1  casenumber2 src2 status2 pen_dt2 
-1   2345  POPDB CLOSED 2012-01-01 245   JCARE CLOSED 2012-01-02 
-1   235   POPDB CLOSED 2012-01-01 NULL  NULL NULL NULL 
1244  2   POPDB CLOSED 2012-01-01 23   JCARE CLOSED 2012-05-02 
273   2345  POPDB CLOSED 2012-03-04 NULL  NULL NULL NULL