我有以下示例中的数据，每个单元格都是事件数。
| 29/09/2016 | 06/10/2016 | 13/10/2016 | Today | Hour |
|------------|------------|------------|-------|------|
| 470 | 649 | 652 | 140 | 0 |
| 159 | 322 | 354 | 925 | 1 |
| 874 | 399 | 412 | 168 | 2 |
| 699 | 312 | 69 | 51 | 3 |
| 911 | 908 | 780 | 322 | 4 |
| 756 | 805 | 579 | 161 | 5 |
| 653 | 295 | 900 | 885 | 6 |
| 637 | 699 | 221 | 310 | 7 |
| 227 | 508 | 382 | 655 | 8 |
| 823 | 782 | 437 | 278 | 9 |
| 389 | 34 | 95 | 754 | 10 |
| 229 | 309 | 29 | 945 | 11 |
| 163 | 16 | 593 | 968 | 12 |
| 833 | 953 | 767 | 438 | 13 |
| 153 | 356 | 619 | 741 | 14 |
| 335 | 493 | 511 | 877 | 15 |
| 469 | 353 | 643 | 431 | 16 |
| 610 | 137 | 121 | 647 | 17 |
| 411 | 951 | 674 | 57 | 18 |
| 593 | 854 | 500 | 380 | 19 |
| 16 | 608 | 445 | 889 | 20 |
| 764 | 143 | 734 | 178 | 21 |
| 709 | 845 | 297 | 292 | 22 |
| 894 | 355 | 309 | 98 | 23 |
我需要比较每小时今天的事件数量与过去3周当天的平均值相比 - 同一小时对比std vs Avg + std vs Avg-std。
| Today-Std | Today+Std | Std | Average | Today | Hour |
|-----------|-----------|-----|---------|-------|------|
| -127 | 407 | 267 | 590 | 140 | 0 |
| 654 | 1196 | 271 | 278 | 925 | 1 |
| -105 | 441 | 273 | 562 | 168 | 2 |
| -224 | 326 | 275 | 360 | 51 | 3 |
| 49 | 595 | 273 | 866 | 322 | 4 |
| -106 | 428 | 267 | 713 | 161 | 5 |
| 617 | 1153 | 268 | 616 | 885 | 6 |
| 43 | 577 | 267 | 519 | 310 | 7 |
| 385 | 925 | 270 | 372 | 655 | 8 |
| 2 | 554 | 276 | 681 | 278 | 9 |
| 478 | 1030 | 276 | 173 | 754 | 10 |
| 674 | 1216 | 271 | 189 | 945 | 11 |
| 702 | 1234 | 266 | 257 | 968 | 12 |
| 182 | 694 | 256 | 851 | 438 | 13 |
| 496 | 986 | 245 | 376 | 741 | 14 |
| 630 | 1124 | 247 | 446 | 877 | 15 |
| 172 | 690 | 259 | 488 | 431 | 16 |
| 374 | 920 | 273 | 289 | 647 | 17 |
| -204 | 318 | 261 | 679 | 57 | 18 |
| 118 | 642 | 262 | 649 | 380 | 19 |
| 613 | 1165 | 276 | 356 | 889 | 20 |
| -88 | 444 | 266 | 547 | 178 | 21 |
| 37 | 547 | 255 | 617 | 292 | 22 |
| -168 | 364 | 266 | 519 | 98 | 23 |
最后，我需要创建图表今日与平均线对比平均线+标准与平均标准。
需要帮助才能在Google Bigquery或SQL Server 2012中执行此操作
由于
答案 0 :(得分：1)
以下是BigQuery Legacy SQL。
我觉得这太简单了，所以我不确定这是不是你要找的东西：
SELECT
Today - Std AS Std_Today,
Today + Std AS Today_Std,
Std,
Average,
Today,
HOUR
FROM (
SELECT
INTEGER(ROUND(SQRT(((d1-Average)*(d1-Average) + (d2-Average)*(d2-Average) + (d3-Average)*(d3-Average))/2))) AS Std,
Average , Today, HOUR
FROM (
SELECT
d1, d2, d3, Today, HOUR, INTEGER(ROUND((d1+d2+d3)/3)) AS Average
FROM (
SELECT d1, d2, d3, Today, HOUR FROM
(SELECT 470 AS d1, 649 AS d2, 652 AS d3, 140 AS Today, 0 AS HOUR),(SELECT 159 AS d1, 322 AS d2, 354 AS d3, 925 AS Today, 1 AS HOUR),
(SELECT 874 AS d1, 399 AS d2, 412 AS d3, 168 AS Today, 2 AS HOUR),(SELECT 699 AS d1, 312 AS d2, 69 AS d3, 51 AS Today, 3 AS HOUR),
(SELECT 911 AS d1, 908 AS d2, 780 AS d3, 322 AS Today, 4 AS HOUR),(SELECT 756 AS d1, 805 AS d2, 579 AS d3, 161 AS Today, 5 AS HOUR),
(SELECT 653 AS d1, 295 AS d2, 900 AS d3, 885 AS Today, 6 AS HOUR),(SELECT 637 AS d1, 699 AS d2, 221 AS d3, 310 AS Today, 7 AS HOUR),
(SELECT 227 AS d1, 508 AS d2, 382 AS d3, 655 AS Today, 8 AS HOUR),(SELECT 823 AS d1, 782 AS d2, 437 AS d3, 278 AS Today, 9 AS HOUR),
(SELECT 389 AS d1, 34 AS d2, 95 AS d3, 754 AS Today, 10 AS HOUR),(SELECT 229 AS d1, 309 AS d2, 29 AS d3, 945 AS Today, 11 AS HOUR),
(SELECT 163 AS d1, 16 AS d2, 593 AS d3, 968 AS Today, 12 AS HOUR),(SELECT 833 AS d1, 953 AS d2, 767 AS d3, 438 AS Today, 13 AS HOUR),
(SELECT 153 AS d1, 356 AS d2, 619 AS d3, 741 AS Today, 14 AS HOUR),(SELECT 335 AS d1, 493 AS d2, 511 AS d3, 877 AS Today, 15 AS HOUR),
(SELECT 469 AS d1, 353 AS d2, 643 AS d3, 431 AS Today, 16 AS HOUR),(SELECT 610 AS d1, 137 AS d2, 121 AS d3, 647 AS Today, 17 AS HOUR),
(SELECT 411 AS d1, 951 AS d2, 674 AS d3, 57 AS Today, 18 AS HOUR),(SELECT 593 AS d1, 854 AS d2, 500 AS d3, 380 AS Today, 19 AS HOUR),
(SELECT 16 AS d1, 608 AS d2, 445 AS d3, 889 AS Today, 20 AS HOUR),(SELECT 764 AS d1, 143 AS d2, 734 AS d3, 178 AS Today, 21 AS HOUR),
(SELECT 709 AS d1, 845 AS d2, 297 AS d3, 292 AS Today, 22 AS HOUR),(SELECT 894 AS d1, 355 AS d2, 309 AS d3, 98 AS Today, 23 AS HOUR)
) as YourTable
)
)
可以轻松修改（如果需要）BigQuery Standard SQL（请参阅Enabling Standard SQL和Migrating from legacy SQL）
从另一方面......
您可以在BigQuery Standard SQL中使用更优雅的代码
WITH YourTable AS (
SELECT 470 AS d1, 649 AS d2, 652 AS d3, 140 AS Today, 0 AS Hour UNION ALL SELECT 159 AS d1, 322 AS d2, 354 AS d3, 925 AS Today, 1 AS Hour UNION ALL
SELECT 874 AS d1, 399 AS d2, 412 AS d3, 168 AS Today, 2 AS Hour UNION ALL SELECT 699 AS d1, 312 AS d2, 69 AS d3, 51 AS Today, 3 AS Hour UNION ALL
SELECT 911 AS d1, 908 AS d2, 780 AS d3, 322 AS Today, 4 AS Hour UNION ALL SELECT 756 AS d1, 805 AS d2, 579 AS d3, 161 AS Today, 5 AS Hour UNION ALL
SELECT 653 AS d1, 295 AS d2, 900 AS d3, 885 AS Today, 6 AS Hour UNION ALL SELECT 637 AS d1, 699 AS d2, 221 AS d3, 310 AS Today, 7 AS Hour UNION ALL
SELECT 227 AS d1, 508 AS d2, 382 AS d3, 655 AS Today, 8 AS Hour UNION ALL SELECT 823 AS d1, 782 AS d2, 437 AS d3, 278 AS Today, 9 AS Hour UNION ALL
SELECT 389 AS d1, 34 AS d2, 95 AS d3, 754 AS Today, 10 AS Hour UNION ALL SELECT 229 AS d1, 309 AS d2, 29 AS d3, 945 AS Today, 11 AS Hour UNION ALL
SELECT 163 AS d1, 16 AS d2, 593 AS d3, 968 AS Today, 12 AS Hour UNION ALL SELECT 833 AS d1, 953 AS d2, 767 AS d3, 438 AS Today, 13 AS Hour UNION ALL
SELECT 153 AS d1, 356 AS d2, 619 AS d3, 741 AS Today, 14 AS Hour UNION ALL SELECT 335 AS d1, 493 AS d2, 511 AS d3, 877 AS Today, 15 AS Hour UNION ALL
SELECT 469 AS d1, 353 AS d2, 643 AS d3, 431 AS Today, 16 AS Hour UNION ALL SELECT 610 AS d1, 137 AS d2, 121 AS d3, 647 AS Today, 17 AS Hour UNION ALL
SELECT 411 AS d1, 951 AS d2, 674 AS d3, 57 AS Today, 18 AS Hour UNION ALL SELECT 593 AS d1, 854 AS d2, 500 AS d3, 380 AS Today, 19 AS Hour UNION ALL
SELECT 16 AS d1, 608 AS d2, 445 AS d3, 889 AS Today, 20 AS Hour UNION ALL SELECT 764 AS d1, 143 AS d2, 734 AS d3, 178 AS Today, 21 AS Hour UNION ALL
SELECT 709 AS d1, 845 AS d2, 297 AS d3, 292 AS Today, 22 AS Hour UNION ALL SELECT 894 AS d1, 355 AS d2, 309 AS d3, 98 AS Today, 23 AS Hour
),
tempTable AS (
SELECT
(SELECT CAST(STDdev(d) AS INT64) FROM UNNEST([d1, d2, d3]) AS d) AS Std,
(SELECT CAST(AVG(d) AS INT64) FROM UNNEST([d1, d2, d3]) AS d) AS Average
, Today, Hour
FROM YourTable
)
SELECT
Today - Std AS Std_Today,
Today + Std AS Today_Std,
Std,
Average,
Today,
Hour
FROM tempTable
ORDER BY Hour
哪（足够有趣）......
将我带回Legacy SQL版本（但与标准版1类似）
SELECT
Today - Std AS Std_Today,
Today + Std AS Today_Std,
Std,
Average,
Today,
HOUR
FROM (
SELECT
INTEGER(STDDEV(d)) AS Std,
INTEGER(AVG(d)) AS Average,
Today, HOUR
FROM (
SELECT Today, HOUR, SPLIT(CONCAT(STRING(d1), ",", STRING(d2), ",", STRING(d3))) AS d
FROM (
SELECT d1, d2, d3, Today, HOUR FROM
(SELECT 470 AS d1, 649 AS d2, 652 AS d3, 140 AS Today, 0 AS HOUR),(SELECT 159 AS d1, 322 AS d2, 354 AS d3, 925 AS Today, 1 AS HOUR),
(SELECT 874 AS d1, 399 AS d2, 412 AS d3, 168 AS Today, 2 AS HOUR),(SELECT 699 AS d1, 312 AS d2, 69 AS d3, 51 AS Today, 3 AS HOUR),
(SELECT 911 AS d1, 908 AS d2, 780 AS d3, 322 AS Today, 4 AS HOUR),(SELECT 756 AS d1, 805 AS d2, 579 AS d3, 161 AS Today, 5 AS HOUR),
(SELECT 653 AS d1, 295 AS d2, 900 AS d3, 885 AS Today, 6 AS HOUR),(SELECT 637 AS d1, 699 AS d2, 221 AS d3, 310 AS Today, 7 AS HOUR),
(SELECT 227 AS d1, 508 AS d2, 382 AS d3, 655 AS Today, 8 AS HOUR),(SELECT 823 AS d1, 782 AS d2, 437 AS d3, 278 AS Today, 9 AS HOUR),
(SELECT 389 AS d1, 34 AS d2, 95 AS d3, 754 AS Today, 10 AS HOUR),(SELECT 229 AS d1, 309 AS d2, 29 AS d3, 945 AS Today, 11 AS HOUR),
(SELECT 163 AS d1, 16 AS d2, 593 AS d3, 968 AS Today, 12 AS HOUR),(SELECT 833 AS d1, 953 AS d2, 767 AS d3, 438 AS Today, 13 AS HOUR),
(SELECT 153 AS d1, 356 AS d2, 619 AS d3, 741 AS Today, 14 AS HOUR),(SELECT 335 AS d1, 493 AS d2, 511 AS d3, 877 AS Today, 15 AS HOUR),
(SELECT 469 AS d1, 353 AS d2, 643 AS d3, 431 AS Today, 16 AS HOUR),(SELECT 610 AS d1, 137 AS d2, 121 AS d3, 647 AS Today, 17 AS HOUR),
(SELECT 411 AS d1, 951 AS d2, 674 AS d3, 57 AS Today, 18 AS HOUR),(SELECT 593 AS d1, 854 AS d2, 500 AS d3, 380 AS Today, 19 AS HOUR),
(SELECT 16 AS d1, 608 AS d2, 445 AS d3, 889 AS Today, 20 AS HOUR),(SELECT 764 AS d1, 143 AS d2, 734 AS d3, 178 AS Today, 21 AS HOUR),
(SELECT 709 AS d1, 845 AS d2, 297 AS d3, 292 AS Today, 22 AS HOUR),(SELECT 894 AS d1, 355 AS d2, 309 AS d3, 98 AS Today, 23 AS HOUR)
) AS YourTable
)
GROUP BY Today, HOUR
)
答案 1 :(得分：1)
下面是对我原来答案的评论 - 原始问题中的数据模式不正确，但如下所示：
events hour date
649 0 06/10/2016
652 0 13/10/2016
140 0 20/10/2016
470 0 29/09/2016
322 1 06/10/2016
354 1 13/10/2016
925 1 20/10/2016
159 1 29/09/2016
399 2 06/10/2016
因此假设日期存储为字符串，其中包含＆＃34; dd / mm / yyyy＆＃34; format - 以下是BigQuery Standard SQL的查询（请参阅Enabling Standard SQL和Migrating from legacy SQL）
SELECT
Date, Hour, Events, Average, Std,
events - Std AS Std_Events,
events + Std AS Events_Std
FROM (
SELECT
x.DATE, x.hour, x.events,
CAST(STDDEV(z.events) AS INT64) AS Std,
CAST(AVG(z.events) AS INT64) Average
FROM yourTable AS x
JOIN (SELECT '20/10/2016' AS xDay) AS y -- use this line to set specific day to treat as today
--JOIN (SELECT FORMAT_DATE('%d/%m/%Y', DATE_SUB(CURRENT_DATE(), INTERVAL 2 DAY)) AS xDay) AS y -- use this line to do analysis for yesterday
ON x.date = xDay
JOIN yourTable as z
ON MOD(DATE_DIFF(PARSE_DATE('%d/%m/%Y', xDay), PARSE_DATE('%d/%m/%Y', z.Date), DAY), 7) = 0
AND DIV(DATE_DIFF(PARSE_DATE('%d/%m/%Y', xDay), PARSE_DATE('%d/%m/%Y', z.Date), DAY), 7) BETWEEN 1 AND 3
AND z.hour = x.hour
GROUP BY 1, 2, 3
)
ORDER BY hour
你可以转换
JOIN (SELECT '20/10/2016' AS xDay) AS y
您需要在一次运行中分析几个日期 - 例如
JOIN (SELECT '19/10/2016' AS xDay UNION ALL
SELECT '20/10/2016' AS xDay) AS y