我正在通过Google BigQuery连接视图。我有一个视图events_with_total_users，该视图连接了一个事件表和一个用户表，因此它包含多个事件，例如“个人资料图片已更新”或“用户电子邮件已更新”，以及用户的ID，user_group和国家/地区以及用户的数量类别和国家/地区的组合（并非每个用户都有一个事件）。
+---------+----------------------------+---------+------------+-----------+
| user_id | event | country | user_group | user_base |
+---------+----------------------------+---------+------------+-----------+
| 1 | Profile picture change | US | free | 798 |
| 7 | Profile picture change | US | free | 798 |
| 2 | Profile description change | CA | premium | 1239 |
| 2 | Language change | CA | premium | 1239 |
| 2 | Email address change | CA | premium | 1239 |
| 3 | Profile picture change | CA | free | 1098 |
| 4 | Email address change | IT | free | 1790 |
| 5 | Language change | DE | premium | 890 |
| 6 | Profile picture change | DE | free | 2673 |
| 6 | Email address change | AT | premium | 745 |
+---------+----------------------------+---------+------------+-----------+
我想创建一个Superset指标，该指标显示国家/地区级别的相对变化数，但能够过滤事件和user_group。
示例：
Percentage of free users that changed their profile picture in the US: 2 / 798 = 0.22%
SQL查询以可视化我在SQL Lab中基于源表events_with_total_users在国家/地区级别创建的百分比：
WITH totals_per_user_group_and_country
AS
(
SELECT country,
user_group,
avg(total_users) total_users
FROM `events_with_total_users`
GROUP BY 1,
2),
totals_per_country
AS
(
SELECT country,
sum(total_users) total_users
FROM totals_per_category_and_country
GROUP BY 1)
SELECT a.country,
round(count(DISTINCT id)/avg(b.total_users)*100,2) ratio
FROM `events_with_total_users` a
JOIN totals_per_country b
USING (country)
GROUP BY 1
当我在仪表板上添加此查询的条形图并基于events_with_total_users表创建过滤器框外观时，该过滤器不会影响我的条形图。我希望Superset可以过滤父源。是否可以过滤user_group和event以获得正确的百分比？