计算 SQL 中两列的百分比 - 使用 BigQuery

Question

我正在尝试计算犯罪状态（clearance_status）占每个类别总犯罪（primary_type）的百分比，但我无法正确输出。

这是我的初始查询：

SELECT primary_type, 
       clearance_status, 
       round(count(clearance_status)/(SELECT count(primary_type) from `bigquery-publicdata.austin_crime.crime`)*100,2) as percentage
FROM `bigquery-public-data.austin_crime.crime`
WHERE primary_type = "Theft: BOV"
GROUP BY primary_type, clearance_status

result:

Row primary_type    clearance_status     percentage 
1   Theft: BOV.     Not cleared          8.59
2   Theft: BOV.     null.                0.0
3   Theft: BOV      Cleared by Arrest.   0.21
4   Theft: BOV      Cleared by Exception 0.03

但是查看 COUNT(primary_type) 和 COUNT(clerance_status) 的结果

SELECT primary_type, count(primary_type) as count
FROM `bigquery-public-data.austin_crime.crime`
WHERE primary_type = "Theft: BOV"
GROUP BY primary_type

result:

Row primary_type    count   
1   Theft: BOV.     10545

SELECT clearance_status, count(clearance_status) as count_1
FROM `bigquery-public-data.austin_crime.crime`
WHERE primary_type = "Theft: BOV"
GROUP BY clearance_status

result:

Row clearance_status     count_1    
1   Not cleared.         10028
2   null                 0
3   Cleared by Arrest.   242
4   Cleared by Exception 30

手动，我应该得到 95% 的“未清除”状态，但我的初始查询只产生 8.59。 使用 over() 给出相同的输出。

Answer 1

SELECT
  primary_type,
  clearance_status,
  COUNT(*) AS clearance_status_count,
  (SUM(COUNT(*)) OVER(PARTITION BY primary_type)) AS total,
  COUNT(*) / (SUM(COUNT(*)) OVER(PARTITION BY primary_type)) AS ratio
FROM
  `bigquery-public-data.austin_crime.crime`
GROUP BY
  primary_type,
  clearance_status
ORDER BY
  primary_type,
  clearance_status;

此查询适用于不同的primary_types 。 我包括计数和总数以表明该比率是正确的。

类似的问题在这里： 使用 GROUP BY 计算组的百分比