Question

我有一系列数据值。我想根据值将其分为多个范围，例如3个范围。例如：假设我的数据值看起来像这样

我的最终目标是能够从该集合中巧妙地确定低-中-高-意思是，我需要代码来找出哪个将具有2个良好的边界值-例如<2600，> = 2600到3000， > = 3000。像K-means聚类之类的...值的集合将始终更改，甚至可能在10或100s，而不是这里的1000s。

The attached image show the distribution of the data points across dates on the x axis. Visually it becomes clear how the data is distributed. I want to be able to do the same via SQL

SQL中的

NTILE分成相等的存储区，因此不符合我的要求。有什么想法吗？

Answer 1

我不知道我是否了解您想要的内容，但是如果要将一组数据划分为3组并找到每个组的范围，我认为以下内容将有助于使用NTILE。

首先，我将数据添加到内存表中进行测试，如下所示：-

declare @dataSet table ([value] int)
insert into @dataSet values 
(1669 ),(2653 ),(2662 ),(2669 ),(2701 ),(2711 ),(2748 ),(2770 ),(2821 ),(2832 ),(2848 ),(2850 ),(2855 ),
(2859 ),(2868 ),(2893 ),(2908 ),(2922 ),(2939 ),(2968 ),(2992 ),(3005 ),(3025 ),(3026 ),(3031 ),(3033 ),
(3034 ),(3036 ),(3053 ),(3073 ),(3087 ),(3087 ),(3099 ),(3128 ),(3128 ),(3131 ),(3133 ),(3140 ),(3174 ),
(3182 ),(3182 ),(3189 ),(3204 ),(3204 ),(3214 ),(3219 ),(3224 ),(3225 ),(3236 ),(3244 ),(3272 ),(3295 ),
(3295 ),(3296 ),(3299 ),(3305 ),(3305 ),(3402 )

然后将尝试获取范围。

    ;with cte as (
    select NTILE(3) over (order by [value]) [group],[value] from @dataSet
    )
    select [group],min([value]) [LowerRange],max([value]) [HigherRange] From cte group by [group]

分组是为了获得每个组的上限和下限

结果如下：-

group   LowerRange  HigherRange
=====   ==========  ==========
1       1669        2968
2       2992        3174
3       3182        3402

希望这会有所帮助。

根据值将数据分为多个范围

1 个答案: