Google 大查询 - Firebase 分析 - 屏幕视图的封闭漏斗（参数）答案

【问题标题】：Google big query - Firebase analytics - Closed funnel for screen views (parameters)Google 大查询 - Firebase 分析 - 屏幕视图的封闭漏斗（参数）
【发布时间】：2019-02-22 08:04:07
【问题描述】：

我想为我的 X 屏幕视图获得一个封闭的漏斗，这是事件 screen_view 的参数

我找到了这个非常好的教程 - https://medium.com/firebase-developers/how-do-i-create-a-closed-funnel-in-google-analytics-for-firebase-using-bigquery-6eb2645917e1，但它仅适用于带有事件的封闭漏斗。

我想得到这个：

event_name   event_param    count_users
screen_view  screen_name_1  100
screen_view  screen_name_2  50
screen_view  screen_name_3  20
screen_view  screen_name_4  5

我尝试将教程中提供的代码更改为事件参数，但我到了不知道下一步该做什么的地步。

SELECT *,
IF (value.string_value = "screen_name1", user_pseudo_id, NULL) as funnel_1, 
IF (value.string_value = "screen_name1" AND next_event = "screen_name2", user_pseudo_id, NULL) AS funnel_2
FROM (
  SELECT p.value.string_value, user_pseudo_id , event_timestamp,
  LEAD(p.value.string_value, 1) OVER (PARTITION BY user_pseudo_id ORDER BY event_timestamp) AS next_event
  FROM `ProjectName.analytics_XX.events_20190119` as t1, UNNEST(event_params) as p
  WHERE (p.value.string_value = "screen_name1" OR p.value.string_value = "screen_name2")
  ORDER BY 2,3
  LIMIT 100
)

感谢您的帮助！

【问题讨论】：

有人知道该怎么做吗？

标签： sql firebase google-bigquery firebase-analytics

【解决方案1】：

我找到了解决办法：

    SELECT COUNT(DISTINCT funnel_1) as f1_users, COUNT(DISTINCT funnel_2) as f2_users FROM (
  SELECT *,
  IF (param.value.string_value = "screen_name1", user_pseudo_id, NULL) AS funnel_1,
  IF (param.value.string_value = "screen_name1" AND next_screen = "screen_name2", user_pseudo_id, NULL) AS funnel_2
  FROM (
    SELECT TIMESTAMP_MICROS(event_timestamp), param, user_pseudo_id,
      LEAD(param.value.string_value, 1) OVER (PARTITION BY user_pseudo_id ORDER BY event_timestamp) as next_screen
    FROM `ProjectName.analytics_XX.events_20190119`, unnest(event_params) as param
    WHERE 
      event_name = "screen_view" and 
      param.value.string_value IN ("screen_name1", "screen_name2")
      AND _TABLE_SUFFIX BETWEEN '20190205' AND '20190312'
  )
)

【讨论】：