why not output the mse in the result? for performance?

[zachzh]

No description provided.

[benfred]

MSE can be pretty easily calculated at run time - and seems like it could be a helpful diagnostic to check convergence. I'll see about adding sometime soon

[zachzh]

Yes, it really help. Hope to see that :)

[benfred]

I've added a function to calculate MSE here da1a7fa

I've left off by default since there is still a small performance penalty for calculating this, but if you pass 'calculate_training_loss=True' to alternating_least_squares this will calculate the loss at each iteration and log it out