按用户 ID 分组的 BigQuery Firebase 事件日志答案

【问题标题】：BigQuery Firebase Event Logs Grouped By User Id按用户 ID 分组的 BigQuery Firebase 事件日志
【发布时间】：2018-07-30 16:21:04
【问题描述】：

问题：我想为我的 android 用户提取按用户 ID 分组并按发生时间排序的 Firebase 事件数据流。我在下面创建了两个脚本，但不幸的是，在这两个脚本中，我似乎都无法正确完成最后一部分，即首先成功地将所有 app_instance_ids 分组，而不管时间戳如何。我应该考虑改用不同的 user_ids 吗？

尝试 1 失败：

SELECT
  d.userid,
  c.ev_timestamp,
  c.ev_name
FROM (SELECT 
        user_dim.app_info.app_instance_id as userid
      FROM `firebase-analytics-sample-data.ios_dataset.app_events_*`, UNNEST(event_dim) AS event
      WHERE _TABLE_SUFFIX BETWEEN '20160601' AND '20160603'
      AND user_dim.first_open_timestamp_micros BETWEEN 1464789600000000 AND 1464962400000000
      GROUP BY 1) AS d
  LEFT JOIN (SELECT user_dim.app_info.app_instance_id as userid,
            event.timestamp_micros as ev_timestamp,
            event.name as ev_name
            FROM `firebase-analytics-sample-data.ios_dataset.app_events_*`, UNNEST(event_dim) AS event
            WHERE _TABLE_SUFFIX BETWEEN '20160601' AND '20160603'
            AND user_dim.first_open_timestamp_micros BETWEEN 1464789600000000 AND 1464962400000000) AS c
  ON d.userid = c.userid
ORDER BY 2 ASC
LIMIT 1000;

不成功的尝试 2：

SELECT
  d.userid,
  d.ev_timestamp,
  c.ev_name
FROM (SELECT 
        user_dim.app_info.app_instance_id as userid,
        event.timestamp_micros as ev_timestamp
      FROM `firebase-analytics-sample-data.ios_dataset.app_events_*`, UNNEST(event_dim) AS event
      WHERE _TABLE_SUFFIX BETWEEN '20160601' AND '20160603'
      AND user_dim.first_open_timestamp_micros BETWEEN 1464789600000000 AND 1464962400000000
      GROUP BY 1,2
      ORDER BY 2 ASC) AS d
  LEFT JOIN (SELECT user_dim.app_info.app_instance_id as userid,
            event.timestamp_micros as ev_timestamp,
            event.name as ev_name
            FROM `firebase-analytics-sample-data.ios_dataset.app_events_*`, UNNEST(event_dim) AS event
            WHERE _TABLE_SUFFIX BETWEEN '20160601' AND '20160603'
            AND user_dim.first_open_timestamp_micros BETWEEN 1464789600000000 AND 1464962400000000) AS c
  ON d.userid = c.userid AND d.ev_timestamp = c.ev_timestamp
#ORDER BY 2 ASC
LIMIT 1000;

正确答案（Amod 的答案转换为 New Export Schema）：

SELECT user_pseudo_id, event_timestamp, event_name
FROM `xxxx.analytics_xxxx.events_*`
WHERE _TABLE_SUFFIX BETWEEN '20180630' AND '20180702'
AND user_first_touch_timestamp BETWEEN 1530453600000000 AND 1530468000000000
AND platform = "ANDROID"
ORDER BY 1,2 ASC
LIMIT 1000

【问题讨论】：

如果您从查询中删除 c.ev_name 是否会返回正确的顺序？
是的，确实如此。但是我怎么加入它错了，因为我的数据集中需要那个列。
试试这个 - 在第一个脚本中在最后一个 ORDER BY 2 ASC 之后添加 GROUP BY 2（原因 chk 回答了查询 - stackoverflow.com/questions/5355585/…）
试用提供的解决方案？
是的，谢谢 Amod :) 我更新了上面的答案。

标签： firebase google-bigquery firebase-analytics

【解决方案1】：

试试这个

    SELECT user_dim.app_info.app_instance_id as userid, X.timestamp_micros,X.name as ev_name,X
FROM `firebase-analytics-sample-data.ios_dataset.app_events_*`, UNNEST(event_dim) AS X
WHERE _TABLE_SUFFIX BETWEEN '20160601' AND '20160603'
      AND user_dim.first_open_timestamp_micros BETWEEN 1464789600000000 AND 1464962400000000
      and user_dim.app_info.app_instance_id like 'C4%'
      order by 1,2 ASC

或

SELECT user_dim.app_info.app_instance_id as userid, X.timestamp_micros,X.name as ev_name
FROM `firebase-analytics-sample-data.ios_dataset.app_events_*`, UNNEST(event_dim) AS X
WHERE _TABLE_SUFFIX BETWEEN '20160601' AND '20160603'
      AND user_dim.first_open_timestamp_micros BETWEEN 1464789600000000 AND 1464962400000000
      order by 1,2 ASC

【讨论】：