Question

我希望每周从创建日期到当前日期绘制投资的观看次数。 enter image description here

每当有人访问投资时，我都会将其插入到表格中 investment_id，user_id和timestamp(datetime)。

表格如下：

investment_id

investment_id  user_id     Time_stamp
499            233         2015-01-22 09:00:42
499            256         2015-01-21 09:00:42
499            275         2015-01-20 09:00:42
499            233         2015-01-14 09:00:42
499            233         2015-01-14 09:00:42
499            233         2015-01-10 09:00:42
499            273         2015-01-06 09:00:42
499            347         2015-01-02 09:00:42
499            343         2015-01-01 09:00:42
499            344         2015-01-01 09:00:42

因此，对于2015-01-01 00:00:00创建的investment_id 499，结果如下：

Week       Views
1          4
2          3
3          2
4          1

找到了解决方法：

SELECT  (CASE (ceiling(datediff(Time_stamp,"2015-01-01 00:00:00")/7)) WHEN 0 THEN 1 
ELSE (ceiling(datediff(Time_stamp,"2015-01-01 00:00:00")/7)) END) as weeks, 
count(Investment_Id) as viewCount 
from log_table where Investment_Id =499 group by weeks

Answer 1

Needed是这样的临时/虚拟表：

+ ------------- + ---------- + ---------- + ---------- +
| investment_id | week_start | week_end   | weeknumber |
+ ------------- + ---------- + ---------- + ---------- +
| 499           | 2015-01-02 | 2015-01-09 | 1          |
| 499           | 2015-01-09 | 2015-01-16 | 2          |
| 499           | 2015-01-16 | 2015-01-23 | 3          |
| 499           | 2015-01-23 | 2015-01-30 | 4          |
+ ------------- + ---------- + ---------- + ---------- +

这可以通过一种生成系列来实现：

SELECT
    499 investment_id,
    @wstart := @startdate + INTERVAL @wseq * 7 DAY week_start,
    @wend := @wstart + INTERVAL 7 DAY week_end,
    @wseq := @wseq + 1 weeknumber
FROM
    information_schema.collations
CROSS JOIN
    (SELECT @startdate := (SELECT MIN(DATE(time_stamp)) 
     FROM log_table
     WHERE investment_id = 499), @wseq := 0) u
HAVING
    week_start <= CURRENT_DATE

交叉连接是用于启动用户变量，具有足够行的表可以从该用户变量生成系列。 information_schema.collations始终可用，并且有超过200行。

如果将其放在子查询中，则可以在investent_id和week_start与week_end之间的time_stamp上连接具有投资视图的表。

这将导致：

SELECT
    s.weeknumber,
    COUNT(v.user_id) views
FROM
    (
    SELECT
        499 investment_id,
        @wstart := @startdate + INTERVAL @wseq * 7 DAY week_start,
        @wend := @wstart + INTERVAL 7 DAY week_end,
        @wseq := @wseq + 1 weeknumber
    FROM
        information_schema.collations
    CROSS JOIN
        (SELECT @startdate := (SELECT MIN(DATE(start_date))
         FROM log_table
         WHERE investment_id = 499), @wseq := 0) u
    HAVING
        weekstart <= CURRENT_DATE
    ) s
LEFT JOIN
    log_table v
    ON s.investment_id = v.investment_id
    AND v.time_stamp >= s.week_start
    AND v.time_stamp < s.week_end
GROUP BY s.weeknumber

Answer 2

这样的事情？

select week(time_stamp), count(investment_id)
group by time_stamp, investment_id

Answer 3

两个选项：

使用DatePart方法从datetime列获取一周，但是这不会给你预期的结果，因为你的输出结果显示，6日不会与1月1日和2日的日历周相同。
使用＆＃34; custom＆＃34;确定一周是什么的代码，即。第一周是每个月的第1周和第7周等等。

以下两个选项的示例

    SELECT  *
    into #tmp
    FROM    
    (
    select  499 investment_id,            233 user_id,         '2015-01-22 09:00:42' Time_stamp
        union all
    select  499,            256,         '2015-01-21 09:00:42'
        union all
    select  499,            275,         '2015-01-20 09:00:42'
        union all
    select  499,            233,         '2015-01-14 09:00:42'
        union all
    select  499,            233,         '2015-01-14 09:00:42'
        union all
    select  499,            233,         '2015-01-10 09:00:42'
        union all
    select  499,            273,         '2015-01-06 09:00:42'
        union all
    select  499,            347,         '2015-01-02 09:00:42'
        union all
    select  499,            343,         '2015-01-01 09:00:42'
        union all
    select  499,            344,         '2015-01-01 09:00:42'
    ) a


    Select DATEPART(wk,Time_stamp) week , count(investment_id)
    FROM    #tmp
    group by DATEPART(wk,Time_stamp)


SELECT  b.WeekNumber , count(b.investment_id)
FROM    (
    SELECT  a.* , 
        case 
            when a.DayInMonth between 1 and 7 then 1
            when a.DayInMonth between 8 and 14 then 2
            when a.DayInMonth between 16 and 22 then 3
            when a.DayInMonth between 23 and 29 then 4
            else 5
        end as WeekNumber               
    FROM    (
        Select day(Time_stamp) DayInMonth , *
        FROM    #tmp
    ) a
)b
group by b.WeekNumber

每周查看计数

3 个答案: