Fit function for [logistic regression as a neural network](https://whiteviolin.medium.com/implementing-logistic-regression-as-a-neural-network-from-scratch-eff9d9cc98bc) article

fit.py

def fit(self):
    """
    Maths involved -
    z = w.T * x + b
    y_predicted = a = sigmoid(z)
    dw += (1 / m) * x * dz
    db += dz
    Gradient descent -
    w = w - α * dw
    b = b - α * db
    """
    self.J = 0
    self.J_last = 1
    dW = np.zeros(shape=(self.m, 1))
    self.b = 0
    self.W = np.zeros(shape=(self.m, 1))

    while True:
        Z = np.dot(self.W.T, self.X) + self.b
        A = np.array([self.sigmoid(x) for x in Z])
        dZ = A - self.Y
        dW = (1 / self.m) * (np.dot(self.X, dZ.T))
        db = (1 / self.m) * np.sum(dZ)

        self.J = -np.sum(
            np.multiply(self.Y.T, np.array([np.log(x) for x in A.T]))
            + np.multiply(1 - self.Y.T, np.array([np.log(1 - x) for x in A.T]))
        )

        self.W = self.W - self.alpha * dW
        self.b = self.b - self.alpha * db

        if self.debug:
            print(self.J)

        if abs(self.J - self.J_last) < 1e-5:
            break
        else:
            self.J_last = self.J