均值和方差的计算（滑动平均和标准差）

最新推荐文章于 2022-05-15 22:46:05 发布

nwpu061701

最新推荐文章于 2022-05-15 22:46:05 发布

阅读量3.8k

点赞数 1

分类专栏：数学文章标签：统计学数学

本文链接：https://blog.csdn.net/u014250897/article/details/118466381

版权

数学专栏收录该内容

3 篇文章 2 订阅

订阅专栏

假设有一组序列 $x_{1},x_{2},......,x_{n},x_{n+1},......$ ，期望以n为步进，滑动求得每n个数据的平均值和标准差（即，想要求得， $x_{1},x_{2},...,x_{n}$ 的平均值和标准差， $x_{2},...x_{n},x_{n+1}$ 的平均值和标准差，以此类推）。那么，我们期望通过一种递推公式，来减少计算量。

先上结论（注：用方差替代标准差给出结论），后给证明。

$E_{i+1}=E_{i}+\frac{1}{n}\left ( x_{i+n}-x_{i}\right )$

$D_{i+1}=D_{i}+\frac{\left ( E_{i+1}-E_{i} \right ) }{n}\left \{ \left ( x_{i+n}+ x_{i}\right ) \left (n-1\right )- 2nE_{i}+2x_{i} \right \}$

证明过程：

$E_{i}=\frac{1}{n}\left ( x_{i}+x_{i+1}+...+x_{i+n-1} \right )$

$D_{i}=\frac{1}{n}\sum_{j=i}^{i+n-1}\left ( x_{j}-E_{1} \right )^{2}$

$E_{i+1}=\frac{1}{n}\left ( x_{i+1}+...+x_{i+n}+x_{i+n} \right )$

$D_{i+1}=\frac{1}{n}\sum_{j=i+1}^{i+n}\left (x_{j}-E_{i+1} \right )^{2}$

$E_{i+1}-E_{i}=\frac{1}{n}\left ( x_{i+n}-x_{i}\right )$

$D_{i+1}-D_{i}=\frac{1}{n}\left [ \sum_{j=i+1}^{i+n}\left ( x_{j}-E_{i+1} \right )^{2}-\sum_{j=i}^{i+n-1}\left ( x_{j}-E_{i} \right )^{2}\right ]$

$D_{i+1}-D_{i}=\frac{1}{n}\left [ \sum_{j=i+1}^{i+n-1}\left ( x_{j}-E_{i+1} \right )^{2}+\left ( x_{i+n}-E_{i+1} \right )^{2}-\sum_{j=i+1}^{i+n-1}\left ( x_{j}-E_{i} \right )^{2}-\left ( x_{i}-E_{i} \right )^{2}\right ]$

$D_{i+1}-D_{i}=\frac{1}{n}\left [ \left ( x_{i+n}-E_{i+1} \right )^{2}-\left ( x_{i}-E_{i} \right )^{2}+\sum_{j=i+1}^{i+n-1}\left ( x_{j}-E_{i+1} \right )^{2}-\sum_{j=i+1}^{i+n-1}\left ( x_{j}-E_{i} \right )^{2}\right ]$

$D_{i+1}-D_{i}=\frac{1}{n}\left [ \left ( x_{i+n}-E_{i+1}+ x_{i}-E_{i} \right ) \left ( x_{i+n}-E_{i+1}-x_{i}+E_{i} \right )+\sum_{j=i+1}^{i+n-1}\left ( x_{j}-E_{i+1}+ x_{j}-E_{i} \right ) \left (x_{j}-E_{i+1}- x_{j}+E_{i} \right )\right ]$

$D_{i+1}-D_{i}=\frac{1}{n}\left [ \left ( x_{i+n}+ x_{i}-E_{i+1}-E_{i} \right ) \left ( x_{i+n}-x_{i}-E_{i+1}+E_{i} \right )+\sum_{j=i+1}^{i+n-1}\left ( 2x_{j}-E_{i+1}-E_{i} \right ) \left (E_{i}-E_{i+1} \right )\right ]$

$D_{i+1}-D_{i}=\frac{1}{n}\left [ \left ( x_{i+n}+ x_{i}-E_{i+1}-E_{i} \right ) \left ( n\left ( E_{i+1}-E_{i} \right )-E_{i+1}+E_{i} \right )+\sum_{j=i+1}^{i+n-1}\left ( 2x_{j} \right ) \left (E_{i}-E_{i+1} \right ) +\sum_{j=i+1}^{i+n-1}\left ( -E_{i+1}-E_{i} \right ) \left (E_{i}-E_{i+1} \right ) \right ]$

$D_{i+1}-D_{i}=\frac{1}{n}\left [ \left ( x_{i+n}+ x_{i}-E_{i+1}-E_{i} \right ) \left ( \left (n-1\right )\left ( E_{i+1}-E_{i} \right ) \right )+\left (E_{i}-E_{i+1} \right ) \sum_{j=i+1}^{i+n-1}\left ( 2x_{j} \right ) +\left ( n-1 \right )\left ( E_{i+1}+E_{i} \right ) \left (E_{i+1}-E_{i} \right ) \right ]$

$D_{i+1}-D_{i}=\frac{1}{n}\left [ \left ( x_{i+n}+ x_{i}-E_{i+1}-E_{i} \right ) \left (n-1\right )\left ( E_{i+1}-E_{i} \right ) +\left ( E_{i+1}+E_{i} \right )\left ( n-1 \right ) \left (E_{i+1}-E_{i} \right )+\left (E_{i}-E_{i+1} \right ) \sum_{j=i+1}^{i+n-1}\left ( 2x_{j} \right ) \right ]$

$D_{i+1}-D_{i}=\frac{1}{n}\left [ \left ( x_{i+n}+ x_{i}\right ) \left (n-1\right )\left ( E_{i+1}-E_{i} \right ) +\left (E_{i}-E_{i+1} \right ) \sum_{j=i+1}^{i+n-1}\left ( 2x_{j} \right ) \right ]$

$D_{i+1}-D_{i}=\frac{1}{n}\left \{ \left ( x_{i+n}+ x_{i}\right ) \left (n-1\right )\left ( E_{i+1}-E_{i} \right ) +\left (E_{i}-E_{i+1} \right ) \left [ \sum_{j=i}^{i+n-1}\left ( 2x_{j} \right ) -2x_{i} \right ]\right \}$

$D_{i+1}-D_{i}=\frac{1}{n}\left \{ \left ( x_{i+n}+ x_{i}\right ) \left (n-1\right )\left ( E_{i+1}-E_{i} \right ) +\left (E_{i}-E_{i+1} \right ) \left [ 2nE_{i}-2x_{i} \right ]\right \}$

$D_{i+1}-D_{i}=\frac{\left ( E_{i+1}-E_{i} \right ) }{n}\left \{ \left ( x_{i+n}+ x_{i}\right ) \left (n-1\right )- 2nE_{i}+2x_{i} \right \}$

nwpu061701

关注

1
点赞
踩
14

收藏

觉得还不错? 一键收藏
3
评论
均值和方差的计算（滑动平均和标准差）

假设有一组序列，期望以n为步进，滑动求得每n个数据的平均值和标准差（即，想要求得，的平均值和标准差，的平均值和标准差，以此类推）。那么，我们期望通过一种递推公式，来减少计算量。先上结论（注：用方差替代标准差给出结论），后给证明。证明过程： ...
复制链接

扫一扫