源地址 http://blog.163.com/jianghuchuan@126/blog/static/46094821201102411517509/

数据的标准化处理是指：使得数据的期望为0，方差为1。

zscore

比如：

B =

     1     2     3     4     5     6

     7     8     9    10    11    12

>> B_S=zscore(B)

B_S =

  Columns 1 through 4

   -0.7071   -0.7071   -0.7071   -0.7071

    0.7071    0.7071    0.7071    0.7071

  Columns 5 through 6

   -0.7071   -0.7071

    0.7071    0.7071

std(B)   %标准差（方差的正的平方根）

ans =

  Columns 1 through 4

    4.2426    4.2426    4.2426    4.2426

  Columns 5 through 6

    4.2426    4.2426

var(B)  %方差

ans =

    18    18    18    18    18    18

>> B

B =

     1     2     3     4     5     6

     7     8     9    10    11    12

>> 4.2426*4.2426

ans =

   17.9997

若要求整个矩阵所有元素的均方差，则要使用std2函数：

B =

     1     2     3     4     5     6

     7     8     9    10    11    12

>> std2(B)

ans =

    3.6056

ZSCORE Standardized z score.

    Z = ZSCORE(X) returns a centered, scaled version of X, the same size as X.

    For vector input X, Z is the vector of z-scores (X-MEAN(X)) ./ STD(X). For

    matrix X, z-scores are computed using the mean and standard deviation

    along each column of X.  For higher-dimensional arrays, z-scores are

    computed using the mean and standard deviation along the first

    non-singleton dimension.

 

    The columns of Z have sample mean zero and sample standard deviation one

    (unless a column of X is constant, in which case that column of Z is

    constant at 0).

 

    [Z,MU,SIGMA] = ZSCORE(X) also returns MEAN(X) in MU and STD(X) in SIGMA.

 

    [...] = ZSCORE(X,1) normalizes X using STD(X,1), i.e., by computing the

    standard deviation(s) using N rather than N-1, where N is the length of  %这个很重要（无偏估计问题！！）
    the dimension along which ZSCORE works.  ZSCORE(X,0) is the same as

    ZSCORE(X).

 

    [...] = ZSCORE(X,FLAG,DIM) standardizes X by working along the dimension

    DIM of X. Pass in FLAG==0 to use the default normalization by N-1, or 1

    to use N.