You use w_reg(lamda_N). y_hat_n is your prediction on data point n after learning on all the data.
Quote:
Originally Posted by cbmachine
I am a bit confused about calculating the value for y_hat_n. To calculate this value should I calculate w_reg(lamda_N-1) for N-1 data points and then calculate y_hat_n using X(n)*w_reg(lamda_N-1). Or should it be calculated using w_reg(lamda_N) i.e. w_reg calculated over all N points?
Any input will be highly appreciated
|