如何在Python中对多变量对数正态分布进行采样?

时间:2015-08-12 23:37:04

标签: python numpy statistics scipy probability

使用Python,我如何从多变量对数正态分布中采样数据?例如,对于多变量法线,有两种选择。假设我们有一个3 x 3协方差矩阵和一个三维平均向量mu。

# Method 1
sample = np.random.multivariate_normal(mu, covariance)

# Method 2
L = np.linalg.cholesky(covariance)
sample = L.dot(np.random.randn(3)) + mu

我找到了numpy的numpy.random.lognormal,但这似乎只适用于单变量样本。我也注意到scipy的scipy.stats.lognorm。这似乎有多变量样本的潜力。但是,我无法弄清楚如何做到这一点。

1 个答案:

答案 0 :(得分:2)

多变量对数正态分布随机变量Rv应具有此属性:log(Rv)应遵循正态分布。因此,问题实际上只是生成多元正态分布的随机变量并np.exp

In [1]: import numpy.random as nr

In [2]: cov = np.array([[1.0, 0.2, 0.3,],
                        [0.2, 1.0, 0.3,],
                        [0.3, 0.3, 1.0]])

In [3]: mu  = np.log([0.3, 0.4, 0.5])

In [4]: mvn = nr.multivariate_normal(mu, cov, size=5)

In [5]: mvn   # This is multivariate normal
Out[5]:
array([[-1.36808854, -1.32562291, -1.9706876 ],
       [-2.13119289,  1.28146425,  0.66000019],
       [-2.82590272, -1.22500654, -0.32635701],
       [-0.4967589 , -0.34469589, -2.04084115],
       [-0.85590235, -1.27133544, -0.70959595]])

In [6]: mvln = np.exp(mvn)

In [7]: mvln   # This is multivariate log-normal
Out[7]:
array([[ 0.25459314,  0.26563744,  0.139361  ],
       [ 0.11869562,  3.60190996,  1.9347927 ],
       [ 0.05925514,  0.29375578,  0.72154754],
       [ 0.60849968,  0.70843576,  0.12991938],
       [ 0.42489961,  0.28045684,  0.49184289]])