CODEWITHSUNDEEP
pythonmachine-learningscipyregressionlinear-regression
GensaGames
GensaGames

Reputation: 5788

Linear Regression Gradient

I have very basic sample of Linear Regression. Implementation below (without Regularization)

class Learning:

    def assume(self, weights, x):
        return np.dot(x, np.transpose(weights))

    def cost(self, weights, x, y, lam):
        predict = self.assume(weights, x) \
            .reshape(len(x), 1)

        val = np.sum(np.square(predict - y), axis=0)
        assert val is not None

        assert val.shape == (1,)
        return val[0] / 2 * len(x)

    def grad(self, weights, x, y, lam):
        predict = self.assume(weights, x)\
            .reshape(len(x), 1)

        val = np.sum(np.multiply(
            x, (predict - y)), axis=0)
        assert val is not None

        assert val.shape == weights.shape
        return val / len(x)

And I want to check gradient, that it works, with scipy.optimize. 

learn = Learning()
INPUTS = np.array([[1, 2],
          [1, 3],
          [1, 6]])
OUTPUTS = np.array([[3], [5], [11]])
WEIGHTS = np.array([1, 1])

t_check_grad = scipy.optimize.check_grad(
    learn.cost, learn.grad, WEIGHTS,INPUTS, OUTPUTS, 0)
print(t_check_grad)
# Output will be 73.2241602235811!!!

I manually checked all computation from start to end. And it's actually right implementatino. But in output I see extremely big difference! What is the reason?

Upvotes: 2

Views: 91

Answers (1)

Sandipan Dey
Sandipan Dey

Reputation: 23101

In your cost function you should return

val[0] / (2 * len(x))

instead of val[0] / 2 * len(x). Then you will have

print(t_check_grad)
# 1.20853633278e-07

Upvotes: 2

Related Questions