在matlab中要计算两个矩阵的相关系数只需调用corr2()即可,但 python中好像没有对等的函数(主要是我没找到),因此根据公式实现,并与matlab的结果对比,发现结果一致。
公式matlab文档那截的:
matlab代码:
a = [[1,2,3];[3,2,1];[1,1,2]];
b = [[2,3,5];[7,3,1];[1,2,3]];
c = [[5,3,1];[1,2,3];[1,2,3]];
d = corr2(a,b);
e = corr2(a,c);
python代码:
import numpy as np
import math
def mean2(x):
y = np.sum(x) / np.size(x)
return y
def corr2(a,b):
a = a - mean2(a)
b = b - mean2(b)
r = (a*b).sum() / math.sqrt((a*a).sum() * (b*b).sum())
return r
a = np.array([[1,2,3],[3,2,1],[1,1,2]])
b = np.array([[2,3,5],[7,3,1],[1,2,3]])
c = np.array([[5,3,1],[1,2,3],[1,2,3]])
d = corr2(a,b)
e = corr2(a,c)