轉置SQL表的最簡潔方法

Question

我如何轉置此表，所以我有兩列，一列顯示付費用戶的比例，另一列顯示免費用戶的比例。 另外，在我的腳本中，有什么方法可以使代碼更簡潔，更像“ SQL”？ 似乎還無法在sqlfiddle中使用select語句創建表，因為我收到一條錯誤消息，指出只能在架構屏幕中創建表。 無論如何，我可以將sql語句嵌入到新表中嗎？

我在sqlfiddle上創建了一個腳本，該腳本可為不同類型的用戶計算一些下載指標： http ://sqlfiddle.com/#!9/79bea4/1

Answer 1

SELECT df.Date,
       SUM(CASE WHEN ad.paying_customer = 'No'
                THEN df.downloads ELSE 0 END) /
       SUM(CASE WHEN ad.paying_customer = 'No' THEN 1 ELSE 0 END) AS `Average Downloads/Free User`,
       SUM(CASE WHEN ad.paying_customer = 'Yes'
                THEN df.downloads ELSE 0 END) /
       SUM(CASE WHEN ad.paying_customer = 'Yes' THEN 1 ELSE 0 END) AS `Average Downloads/Paid User`
FROM
(
    SELECT date,
           user_id,
           SUM(downloads) AS downloads
    FROM download_facts
    GROUP BY date,
             user_id
) df
INNER JOIN user_dimension ud
    ON df.user_id = ud.user_id
INNER JOIN account_dimension ad
    ON ud.account_id = ad.account_id
GROUP BY df.Date

演示在這里：

SQLFiddle

Answer 2

您可以通過將下載總和除以用戶數量來計算平均值。 然后count(distinct)可以獲取每個組中的用戶：

select df.date,
       (sum(df.downloads) /
        count(distinct case when ad.paying_customer = 'No' then df.user_id end)
       ) as avg_free,   
       (sum(df.downloads) /
        count(distinct case when ad.paying_customer = 'Yes' then df.user_id end)
       ) as avg_paying   
from download_facts df left join
     user_dimension ud
     on df.user_id = ud.user_id left join
     account_dimension ad
     on ad.account_id = ud.account_id
group by df.date;