got an error for 11.1.6

[fxj483]

my code for the update weights:

def update_weights(X_, y_, w_, eta):
error = compute_error(y_, X_, w_)
fod = X_.T.dot(error)
w_[1:] += eta * np.dot(x_.T,error)
w_[0] += eta * error.sum(

#answer for 11.1.6
w = np.zeros(1+X.shape[1])
for i in range(50):
print(update_weights(y_train, X_train,w,0.001))

the error I got:

It seems the error is about the training data, but my code for handling the data is the same as the solution provided.

Thank you in advance!

[jsr-p]

hi @fxj483 ,
the problem is this line update_weights(y_train, X_train,w,0.001).
Your update_weights function expects the input (X_, y_, w_, eta) but you insert (y_train, X_train,w,0.001). You will have to switch y_train and X_train :)
The ValueError gets raised in the net_input function where you are trying to multiply a vector of dim (122, ) with a vector of dim (8, ). This is not well defined as to why the value error gets raised.

Whenever you get a ValueError due to something with the dimensionality of your data, check the dimension of your feature matrix and target vector by, e.g. for the feature matrix, writing X.shape.