jsr-p
commented
4 years ago 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.
my code for the update weights:
def updateweights(X, y, w, eta): error = computeerror(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!