如何在Python中将2D列向量转换为1D向量以进行Pearson相关系数计算

ValueError: shapes (1000,1) and (1000,1) not aligned: 1 (dim 1) != 1000 (dim 0)

import numpy as np
from scipy import stats

# 为了可复现性，设置随机数种子
rng = np.random.default_rng(483465834568457)

# 创建两个 (1000, 1) 形状的 NumPy 数组作为示例数据
xhand = rng.random(size=(1000, 1))
xpred = rng.random(size=(1000, 1))

print(f"原始 xhand 形状: {xhand.shape}") # 输出: (1000, 1)
print(f"原始 xpred 形状: {xpred.shape}\n") # 输出: (1000, 1)

# 尝试直接计算，会报错
try:
    correlation_coefficient, p_value = stats.pearsonr(xhand, xpred)
except ValueError as e:
    print(f"直接计算错误: {e}\n")

# 使用 .ravel() 方法进行转换
correlation_coefficient_ravel, p_value_ravel = stats.pearsonr(xhand.ravel(), xpred.ravel())
print(f"使用 .ravel() 转换后的 xhand 形状: {xhand.ravel().shape}")
print(f"Pearson R (ravel): {correlation_coefficient_ravel:.4f}, P-value: {p_value_ravel:.4f}\n")

# 使用 .flatten() 方法进行转换
correlation_coefficient_flatten, p_value_flatten = stats.pearsonr(xhand.flatten(), xpred.flatten())
print(f"使用 .flatten() 转换后的 xhand 形状: {xhand.flatten().shape}")
print(f"Pearson R (flatten): {correlation_coefficient_flatten:.4f}, P-value: {p_value_flatten:.4f}\n")

# 使用 .reshape(-1) 方法进行转换
correlation_coefficient_reshape, p_value_reshape = stats.pearsonr(xhand.reshape(-1), xpred.reshape(-1))
print(f"使用 .reshape(-1) 转换后的 xhand 形状: {xhand.reshape(-1).shape}")
print(f"Pearson R (reshape(-1)): {correlation_coefficient_reshape:.4f}, P-value: {p_value_reshape:.4f}\n")

ValueError: x and y must have length at least 2.

import numpy as np
from scipy import stats

rng = np.random.default_rng(483465834568457)

# 创建两个 (1000, 1) 形状的 NumPy 矩阵作为示例数据
xhand_matrix = np.matrix(rng.random(size=(1000, 1)))
xpred_matrix = np.matrix(rng.random(size=(1000, 1)))

print(f"原始 xhand_matrix 类型: {type(xhand_matrix)}") # 输出: <class 'numpy.matrix'>
print(f"原始 xhand_matrix 形状: {xhand_matrix.shape}\n") # 输出: (1000, 1)

# 尝试直接对 np.matrix 使用 reshape(-1) 后计算，可能仍会报错
try:
    stats.pearsonr(xhand_matrix.reshape(-1), xpred_matrix.reshape(-1))
except ValueError as e:
    print(f"对 np.matrix 使用 reshape(-1) 后计算错误: {e}\n")

# 正确的方法：先转换为 np.ndarray，再展平
xhand_array_flat = np.asarray(xhand_matrix).ravel()
xpred_array_flat = np.asarray(xpred_matrix).ravel()

print(f"转换后 xhand_array_flat 类型: {type(xhand_array_flat)}") # 输出: <class 'numpy.ndarray'>
print(f"转换后 xhand_array_flat 形状: {xhand_array_flat.shape}\n") # 输出: (1000,)

correlation_coefficient_matrix, p_value_matrix = stats.pearsonr(xhand_array_flat, xpred_array_flat)
print(f"Pearson R (np.matrix 转换后): {correlation_coefficient_matrix:.4f}, P-value: {p_value_matrix:.4f}\n")