質問編集履歴
3
saziさんの回答を参考に書いたコードの追記。
title
CHANGED
|
File without changes
|
body
CHANGED
|
@@ -40,4 +40,84 @@
|
|
|
40
40
|
event_date
|
|
41
41
|
)
|
|
42
42
|
```
|
|
43
|
-
user_idはイベント毎に計測されるため、元データでは重複があります。
|
|
43
|
+
user_idはイベント毎に計測されるため、元データでは重複があります。
|
|
44
|
+
|
|
45
|
+
|
|
46
|
+
＜↓saziさんの回答を自分なりに解釈して書いたコード（コピペだとエラーで、saziさんの書き方が高度でどう変えればいいかわかりませんでした。。。）＞
|
|
47
|
+
エラーはないのですが、dau、wau、mauの数字がほぼ同じになってしまいました。
|
|
48
|
+
|
|
49
|
+
```standardsql
|
|
50
|
+
with T as (
|
|
51
|
+
select
|
|
52
|
+
parse_date("%Y%m%d", event_date) event_date
|
|
53
|
+
, user_id
|
|
54
|
+
, event_name
|
|
55
|
+
from
|
|
56
|
+
`sample*`
|
|
57
|
+
where(
|
|
58
|
+
_TABLE_SUFFIX BETWEEN
|
|
59
|
+
"20180707" AND
|
|
60
|
+
FORMAT_DATE("%Y%m%d", DATE_SUB(CURRENT_DATE(), INTERVAL 1 DAY))
|
|
61
|
+
)
|
|
62
|
+
and
|
|
63
|
+
app_info.install_source = 'iTunes'
|
|
64
|
+
and
|
|
65
|
+
date(TIMESTAMP_ADD(timestamp_micros(user_first_touch_timestamp), interval device.time_zone_offset_seconds SECOND)) > "2018-07-06"
|
|
66
|
+
and
|
|
67
|
+
user_id is not null
|
|
68
|
+
limit 10000000
|
|
69
|
+
)
|
|
70
|
+
|
|
71
|
+
, dau as (
|
|
72
|
+
select
|
|
73
|
+
event_date
|
|
74
|
+
,count(distinct(user_id)) as dau_count
|
|
75
|
+
from
|
|
76
|
+
T
|
|
77
|
+
group by
|
|
78
|
+
event_date
|
|
79
|
+
)
|
|
80
|
+
|
|
81
|
+
, wau as (
|
|
82
|
+
select
|
|
83
|
+
event_date
|
|
84
|
+
,count(distinct(user_id)) as wau_count
|
|
85
|
+
from
|
|
86
|
+
T
|
|
87
|
+
where
|
|
88
|
+
event_date between t.event_date and date_add(t.event_date,interval 7 day)
|
|
89
|
+
group by event_date
|
|
90
|
+
)
|
|
91
|
+
|
|
92
|
+
, mau as (
|
|
93
|
+
select
|
|
94
|
+
event_date
|
|
95
|
+
,count(distinct(user_id)) as mau_count
|
|
96
|
+
from
|
|
97
|
+
T
|
|
98
|
+
where
|
|
99
|
+
event_date between t.event_date and date_add(t.event_date,interval 1 month)
|
|
100
|
+
group by event_date
|
|
101
|
+
)
|
|
102
|
+
|
|
103
|
+
, before_formating as (
|
|
104
|
+
select
|
|
105
|
+
dau.event_date
|
|
106
|
+
,dau_count
|
|
107
|
+
,wau_count
|
|
108
|
+
,mau_count
|
|
109
|
+
from
|
|
110
|
+
dau
|
|
111
|
+
left join
|
|
112
|
+
wau
|
|
113
|
+
on
|
|
114
|
+
dau.event_date = wau.event_date
|
|
115
|
+
left join
|
|
116
|
+
mau
|
|
117
|
+
on
|
|
118
|
+
dau.event_date = mau.event_date
|
|
119
|
+
)
|
|
120
|
+
select *
|
|
121
|
+
from before_formating
|
|
122
|
+
order by event_date asc
|
|
123
|
+
```
|
2
user_idの状態に関して追記
title
CHANGED
|
File without changes
|
body
CHANGED
|
@@ -39,4 +39,5 @@
|
|
|
39
39
|
group by
|
|
40
40
|
event_date
|
|
41
41
|
)
|
|
42
|
-
```
|
|
42
|
+
```
|
|
43
|
+
user_idはイベント毎に計測されるため、元データでは重複があります。
|
1
Tテーブルの追加
title
CHANGED
|
File without changes
|
body
CHANGED
|
@@ -10,6 +10,26 @@
|
|
|
10
10
|
何を使ってどう集計すればいいのでしょうか？
|
|
11
11
|
ちなみにDAUは以下のように出しています。
|
|
12
12
|
```standardsql
|
|
13
|
+
with T as (
|
|
14
|
+
select
|
|
15
|
+
parse_date("%Y%m%d", event_date) event_date
|
|
16
|
+
, user_id
|
|
17
|
+
, event_name
|
|
18
|
+
from
|
|
19
|
+
`sample*`
|
|
20
|
+
where(
|
|
21
|
+
_TABLE_SUFFIX BETWEEN
|
|
22
|
+
"20180707" AND
|
|
23
|
+
FORMAT_DATE("%Y%m%d", DATE_SUB(CURRENT_DATE(), INTERVAL 1 DAY))
|
|
24
|
+
)
|
|
25
|
+
and
|
|
26
|
+
app_info.install_source = 'iTunes'
|
|
27
|
+
and
|
|
28
|
+
date(TIMESTAMP_ADD(timestamp_micros(user_first_touch_timestamp), interval device.time_zone_offset_seconds SECOND)) > "2018-07-06"
|
|
29
|
+
and
|
|
30
|
+
user_id is not null
|
|
31
|
+
)
|
|
32
|
+
|
|
13
33
|
,dau as (
|
|
14
34
|
select
|
|
15
35
|
event_date
|