multiply matrix

# выделим в данных признаки и целевой признак
features = data.drop(['Страховые выплаты'], axis=1)
target = data['Страховые выплаты']

# разделим на обучающую и тестовую выборки
train_features, test_features, train_target, test_target = train_test_split(
                                                            features, target, test_size = 0.25,
                                                            random_state=1234)

class LinearRegression:
    def fit(self, train_features, train_target):
        X = np.concatenate((np.ones((train_features.shape[0], 1)), train_features), axis=1)
        y = train_target
        w = (np.linalg.inv(X.T.dot(X)).dot(X.T)).dot(y)
        self.w = w[1:]
        self.w0 = w[0]

    def predict(self, test_features):
        return test_features.dot(self.w) + self.w0

model = LinearRegression()
model.fit(train_features, train_target)
predictions = model.predict(test_features)
print(r2_score(test_target, predictions))

#Теперь сгенерируем случайную матрицу функцией numpy.random.normal()

np.random.seed(100) # фиксируем случайность

P =  np.random.normal(3, 2.5, size=(4, 4))
A = X @ P
A.shape

model.fit(A, train_target)
predictions = model.predict(A)
print(r2_score(test_target, predictions))


---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
/tmp/ipykernel_31/3434024009.py in <module>
----> 1 model.fit(A, train_target)
      2 predictions = model.predict(A)
      3 print(r2_score(test_target, predictions))

/tmp/ipykernel_31/3890212866.py in fit(self, train_features, train_target)
      3         X = np.concatenate((np.ones((train_features.shape[0], 1)), train_features), axis=1)
      4         y = train_target
----> 5         w = (np.linalg.inv(X.T.dot(X)).dot(X.T)).dot(y)
      6         self.w = w[1:]
      7         self.w0 = w[0]

ValueError: shapes (5,4847) and (3635,) not aligned: 4847 (dim 1) != 3635 (dim 0)