ValueError: Unable to coerce to Series, length must be 1: given 506

Question

I'm trying to apply the Simple Linear regression with gradient descent but I got stuck here

I get an error ValueError

ValueError: Unable to coerce to Series, length must be 1: given 506 can anyone help me what it the wrong in this code

X_train_std: standarding value of X, X.shape = (506, 13) X have int values y.shape = (506,1)

class LinearRegressionGD(object):
    def __init__(self, eta = 0.001, n_iter = 20):
        self.eta = eta
        self.n_iter = n_iter

    def fit(self, X_train_std, y):
        self.w_ = np.zeros(1+X_train_std.shape[1])
        self.cost = []

        for i in range(self.n_iter):
            output = self.net_input(X_train_std)
            errors = (y - output)
            self.w_[1:] += self.eta * X_train_std.T.dot(errors)
            self.w_[0] += self.eta * errors.sum()
            cost = (errors**2).sum() /2.0
            self.cost_.append(cost)
        return self

    def net_input(self,X_train_std):
        return np.dot(X_train_std, self.w_[1:]) + self.w_[0]
    
    def predict(self, X_train_std):
        return self.net_input(X_train_std)


lr = LinearRegressionGD()
lr.fit(X_train_std, y)



---------------------------------------------------------------------------
    ValueError                                Traceback (most recent call last)
    
    <ipython-input-24-29cc349fb131> in <module>()
          1 lr = LinearRegressionGD()
    ----> 2 lr.fit(X_train_std, y)
    
    3 frames
    
    /usr/local/lib/python3.6/dist-packages/pandas/core/ops/__init__.py in to_series(right)
        637             if len(left.columns) != len(right):
        638                 raise ValueError(
    --> 639                     msg.format(req_len=len(left.columns), given_len=len(right))
        640                 )
        641             right = left._constructor_sliced(right, index=left.columns)

Answer 1

output is one dimensional while y is two dimensional. That's why you have to reshape the output.

output = self.net_input(X_train_std).reshape(-1, 1)

Also I get no convergence with your weights updates. If you replace it with

self.w_[1:] += self.eta * np.sum(errors * X_train_std, axis=0)

and probably tweak learning rate it should work fine.