构建日期

SQL Server 中的时间序列分析

Maham Faisal Khan

Senior Data Science Content Developer

您将学到

使用日期组成部分
将字符串转换为日期，含时区偏移与无效日期
按时间段过滤、分组与聚合
降采样数据
窗口聚合
升采样
计算累计和与移动平均
查找日期区间重叠

构建一个日期

SELECT
    GETDATE() AS DateTime_LTz,
    GETUTCDATE() AS DateTime_UTC;

SELECT
    SYSDATETIME() AS DateTime2_LTz
    SYSUTCDATETIME() AS DateTime2_UTC;

Results:

DateTime_LTz	DateTime_UTC	DateTime2_LTz	DateTime2_UTC
2019-03-07 21:21:33.670	2019-03-08 02:21:33.670	2019-03-07 21:21:33.6716402	2019-03-08 02:21:33.6716402

拆解日期

DECLARE
    @SomeDate DATETIME2(3) = '2019-03-01 08:17:19.332';

SELECT YEAR(@SomeDate);

SELECT MONTH(@SomeDate);

SELECT DAY(@SomeDate);

YEAR = 2019

MONTH = 3

DAY = 1

用日期组成部分解析

函数

DATEPART()

SELECT
  DATEPART(YEAR, @dt) AS TheYear;

DATENAME()

SELECT
  DATENAME(MONTH, @dt) AS TheMonth;

组成部分

年 / 月 / 日
年内第几天
周几
年内第几周
ISO 年内第几周
分钟 / 秒
毫秒 / 纳秒

日期的加减

DECLARE
    @SomeTime DATETIME2(7) = '1992-07-14 14:49:36.2294852';

SELECT
    DATEADD(DAY, 1, @SomeTime) AS NextDay,
    DATEADD(DAY, -1, @SomeTime) AS PriorDay;

SELECT
    DATEADD(HOUR, -3, DATEADD(DAY, -4, @SomeTime)) AS Minus4Days3Hours;

NextDay	PriorDay
1992-07-15 14:49:36.2294852	1992-07-13 14:49:36.2294852

Minus4Days3Hours
1992-07-10 11:49:36.2294852

比较日期

DECLARE
    @StartTime DATETIME2(7) = '2012-03-01 14:29:36',
    @EndTime DATETIME2(7) = '2012-03-01 18:00:00';

SELECT
    DATEDIFF(SECOND, @StartTime, @EndTime) AS SecondsElapsed,
    DATEDIFF(MINUTE, @StartTime, @EndTime) AS MinutesElapsed,
    DATEDIFF(HOUR, @StartTime, @EndTime) AS HoursElapsed;

SecondsElapsed	MinutesElapsed	HoursElapsed
12624	211	4

日期取整

DECLARE @SomeTime DATETIME2(7) = '2017-05-12 16:25:16.2248991';

SELECT
    DATEADD(DAY, DATEDIFF(DAY, 0, @SomeTime), 0) AS RoundedToDay,
    DATEADD(HOUR, DATEDIFF(HOUR, 0, @SomeTime), 0) AS RoundedToHour,
    DATEADD(MINUTE, DATEDIFF(MINUTE, 0, @SomeTime), 0) AS RoundedToMinute;

RoundedToDay          RoundedToHour         RoundedToMinute
2017-05-12 00:00:00   2017-05-12 16:00:00   2017-05-12 16:25:00

Passons à la pratique !

SQL Server 中的时间序列分析