如何獲得團體

Question

的不同組合，我有一個表像

name_id  disease_id 
-------  ----------  
1   1  
1   2  
2   2  
3   1  
3   3  
4   1  
4   2  
5   2  

我需要找到disease_id的整個表的不同組合。我無法按name_id進行分組，因爲這會產生不明顯的組合（請參閱name_id 1和4）。我不能忽略GROUP BY，它會在所有name_id中生成一個簡單的disease_id列表。

我希望能夠給的答覆是這樣的問題：什麼是羣體內發生的（以及後來也許增加一個計數兒童疾病的獨特組合：

Chickenpox   (10) 
Chickenpox+Measles (2) 
Measles    (5) 
Measles+Mumps  (1) 
etc. 

Answer 1

解決方案有兩個步驟。

1. 列舉每個患者患病的一組疾病;將其存儲在單獨的表中
2. 使用上表列出疾病的不同組合以及患有這些組合的患者數。

現在，對於步驟＃1你需要一個存儲過程，而云爲下：

DECLARE @prv int 
DECLARE @nid int 
DECLARE @dname varchar(100) 
DECLARE @combi varchar(500) 

DECLARE c1 CURSOR FOR 
SELECT name_id, disease_name 
FROM patients 
JOIN diseases ON patients.disease_id = diseases.disease_id 
ORDER BY name_id, disease_name; 

DELETE FROM diseasecombi; 

OPEN c1 
FETCH NEXT FROM c1 INTO @nid, @dname 
SET @prv = @nid 
SET @combi = '' 

WHILE @@FETCH_STATUS = 0 
BEGIN 
    IF @prv <> @nid 
    BEGIN 
     INSERT INTO diseasecombi (name_id, suffers) VALUES (@prv, @combi) 
     SET @prv = @nid 
     SET @combi = '' 
    END 
    IF LEN(@combi) > 0 SET @combi = @combi + ', ' 
    SET @combi = @combi + @dname 
    FETCH NEXT FROM c1 INTO @nid, @dname 
END 
INSERT INTO diseasecombi(name_id, suffers) VALUES (@prv, @combi) 

CLOSE c1 
DEALLOCATE c1 

的存儲過程上面會產生一個表，看起來像

name_id suffers 
------- ------------------------------- 
1  Chickenpox, Mumps 
2  Chickenpox, Fibroids, Measles 
3  Chickenpox, Mumps 
4  Chickenpox, Measles 
5  Chickenpox, Measles 
6  Chickenpox 
7  Rashes 

對於步驟＃2，查詢如下：

SELECT suffers, COUNT(*) AS Patients 
FROM diseasecombi 
GROUP BY suffers 
ORDER BY suffers 

W HICH產生輸出下：

Diseases      Patients 
------------------------------ ---------- 
Chickenpox      1 
Chickenpox, Fibroids, Measles 1 
Chickenpox, Measles    2 
Chickenpox, Mumps    2 
Rashes       1 

Answer 2

你想成爲能夠回答諸如「什麼是羣體內發生的兒童疾病的獨特組合」問題

我的答案是基於以下假設：

1. 疾病_id指水痘，麻疹，腮腺炎等
2. name_id指患者
3. 疾病_id 1指'水痘'和疾病_id 3指'麻疹';兩者都被視爲兒童疾病。 （這僅僅是舉例來說，你可以不同）

基於上述假設，查詢下面產生所需的結果：

SELECT 'Chickenpox' as 'DiseaseName', COUNT(*) as 'Patients' 
FROM (
    SELECT name_id FROM table GROUP BY name_id HAVING COUNT(*) = 1 WHERE disease_id = 1 
    ) d1 
UNION 
SELECT 'Measles' as 'DiseaseName', COUNT(*) as 'Patients' 
FROM (
    SELECT name_id FROM table GROUP BY name_id HAVING COUNT(*) = 1 WHERE disease_id = 3 
    ) d3 
UNION 
SELECT 'Chickenpox + Measles' as 'DiseaseName', COUNT(*) as 'Patients' 
FROM (
    SELECT name_id FROM table GROUP BY name_id HAVING COUNT(*) = 2 WHERE disease_id IN (1,3) 
    ) d1d3 

結果將被作爲下：

DiseaseName   Patients 
-------------------- --------- 
Chickenpox   10 
Measles    5 
Chickenpox+Measels 2