numpy中矩阵无法相乘

Question

在使用numpy和sklearn自主实现逻辑回归的过程中，矩阵无法相乘

```python
import sys
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.datasets import load_breast_cancer
data = load_breast_cancer(return_X_y=True)
X = np.array(data[0])
y = np.array(data[1])

def sigmod(x):
    return 1/(1+pow(np.e,(-x)))

def Logistic_Regression(feature_data,target_data,learning_rate,account):
    m = feature_data.shape[0]
    feature_data = np.hstack((np.full((m,1),1),feature_data))
    m,n = feature_data.shape
    para = np.random.uniform(-1,1,n).reshape(n,1) # n*1
    para = np.mat(para)
    feature_data = np.mat(feature_data)    # m*n
    target_data = np.mat(target_data)
    # =========问题代码，两矩阵阵无法相乘？？？？============#
    d = pd.DataFrame(np.array(feature_data))
    print(d.describe())
    print(type(feature_data), type(para))
    print(feature_data.shape, para.shape)
    print(feature_data @ para)

    Error = (-1/m)*np.sum(np.multiply(target_data,np.log(sigmod(feature_data @ para))) +
                    np.multiply(1-target_data,np.log(1-sigmod(feature_data @ para))))
    count = 1
    error_list = [Error]

    while True:
        grad_vector = (1/m) * [feature_data.T @ (sigmod(feature_data @ para) - target_data)]
        para = para - learning_rate * grad_vector
        Error = (-1 / m) * np.sum(np.multiply(target_data, np.log(sigmod(feature_data @ para))) +
                                np.multiply(1 - target_data, np.log(1 - sigmod(feature_data @ para))))
        error_list.append(Error)
        count = count + 1
        if count == account:
            break
    return para,error_list

para,e = Logistic_Regression(X,y,0.01,100)
plt.plot(e)
plt.show()

###### 无显式报错，但程序中止

###### 试过更改库版本，无果

Answer 1

np包的矩阵相乘有两种你需要逐个分清，dot和multiply。
如果是矩阵的相乘的话需要检查两个矩阵的长和宽是否在每一次的迭代中都符合要求，同时也要避免矩阵中出现None值