sql - 参数 * 只能用于 COUNT(*)

Question

我的表如下所示：

examID action   timeSinceStart
a     write     300
a     answer   2535
b     answer   3834
a     answer   2900
c     write     774
d     edit     3674
c     answer   3489
a     edit     3983
c     answer    300

我寻求以下输出：

examID averageTime
a      2717.5
b      3834
c      1894.5    

也就是说，我只看actions那些answer，我只写了examID一次又一次的专栏average time since start at the point of answering。

是否可以直接执行此操作而无需创建总和列和计数列，然后编写平均公式？

我的“复杂”方法不起作用：

%%bigquery df_average_times
SELECT countTimes.examID, sumTimes.answerSum, countTimes.answerCount, (sumTimes/countTimes) as averageTime
FROM
 ( SELECT examID, timeSinceStart, action, SUM(*) as answerSum
 FROM `some-database.bq.results` 
 GROUP BY examID, timeSinceStart, action
 HAVING action='answer'
 ORDER BY examID) as sumTimes
 INNER JOIN
 ( SELECT examID, timeSinceStart, action, COUNT(*) as answerCount
 FROM `some-database.bq.results` 
 GROUP BY examID, timeSinceStart, action
 HAVING event='answer'
 ORDER BY examID) as countTimes
 ON falseAnswers.questionId = trueAnswers.questionId

错误指出：

ERROR:
 400 Argument * can only be used in COUNT(*) at [3:58]

Answer 1

我认为您需要以下查询-

SELECT examID, AVG(timeSinceStart)
  FROM (SELECT examID, timeSinceStart
          FROM `some-database.bq.results`
         WHERE action='answer') X
 GROUP BY examID;

Answer 2

如果您想要答案的平均时间，您可以使用：

SELECT examID, AVG(timeSinceStart)
FROM `some-database.bq.results`
WHERE action = 'answer' 
GROUP BY examID;

如果你想让这个有条件——比如在同一个查询中获取不同操作的平均值——你可以使用一个CASE表达式：

SELECT examID,
       AVG(CASE WHEN action = 'answer' THEN timeSinceStart END) avg_answer,
       AVG(CASE WHEN action = 'write' THEN timeSinceStart END) as avg_write
FROM `some-database.bq.results`
WHERE action = 'answer' 
GROUP BY examID;