Trying to understand max_iterations for fitting classifier

User 3226 | 3/7/2016, 5:26:01 AM

I am running logistic regression and I set the max_iterations to 25, I got this message "This model may not be optimal. To improve it, consider increasing max_iterations.

Can anyone help to explain to me what does the max_iteration mean? Does it work like ensemble which takes the average of multiple iteration output?

Comments

User 1592 | 3/7/2016, 9:29:28 AM

Hi Yuming, Logistic regression is an iterative algorithm which refines its state in each iteration. Max iterations define how many iterations (passes on the full dataset) will be performed. This warning means that we have detected it will be useful to increase the number if iterations to get improved accuracy.