python - How can I implement multiclass logistic regression from scratch?

Question

I've been trying to implement multiclass logistic regression. Following is my code for two classes.

class LogReg:
    def __init__(self, lr=0.0001, itr=1000):
        self.lr = lr
        self.itr= itr
        self.weight = None
        self.bias = None
    def sigmoid(self, x):
        return 1 / (1 + np.exp(-x))
    def fit(self, X, y):
        entries, features = X.shape
        self.weight = np.zeros(features)
        self.bias = 0
        for _ in range(self.itr):
            thetaX = np.dot(X, self.weight) + self.bias
            h= self.sigmoid(thetaX)
            dw = (1 / entries) * np.dot(X.T, (h - y))
            db = (1 / entries) * np.sum(h - y)
            self.weight -= self.lr * dw
            self.bias -= self.lr * db
    def predict(self, X):
        thetaX= np.dot(X, self.weight) + self.bias
        h = self.sigmoid(thetaX)
        op = [1 if i > 0.5 else 0 for i in h]
        return np.array(op)

It would be great help if someone could explain if (then how) it's possible to code a multiclass classifier in a similar fashion as shown above.

Answer 1

等待大神答复