I'm a newbie in machine learning field. And I need to choose the best model for my classification model, so i use "learning curve" from sk-learn to make selection. I train and plot learning curve on models such:

  • Logistic Regression
  • KNN
  • Gaussian Naive Bayes
  • Decision Tree
  • Random forest
  • Stochastic Gradient Descent
  • Perceptron
  • SVM
  • Linear SVC

Here is detail graphs:

enter image description here

So I don't know how to conclude from above pictures:

  1. How bias vs variance on each model (underestimate/over estimate) ?
  2. Which models need more data to train or need to increase the complex ( increase degree,...)
  3. Which models is good to apply?
  4. Which modes is the best to select?

Please help me clarify these answers relate to these learning curves.

Thanks a lot

  • $\begingroup$ Cross-posted: stackoverflow.com/q/45518709/781723. Please do not post the same question on multiple sites. Each community should have an honest shot at answering without anybody's time being wasted. $\endgroup$ – D.W. Aug 6 '17 at 5:21
  • $\begingroup$ I'm voting to close this question because it was cross-posted. $\endgroup$ – D.W. Aug 6 '17 at 5:22
  • $\begingroup$ Sorry for that. First time I post on stackoverflow(without computer science). And I don't receive any feedback from its link. So I found, stackoverflow have specify are for CS, so I post another question at here. Please don't close my post. I really need the support from the community $\endgroup$ – Jame H Aug 7 '17 at 4:41
  • $\begingroup$ You should pick one site where you want it to appear, and delete the copy on the other site. That said, this question probably needs some editing before it is suitable here. We want you to ask only one question per post. (If you have multiple questions, they can be posted separately.) (continued) $\endgroup$ – D.W. Aug 7 '17 at 5:31
  • $\begingroup$ We also want to know what research and self-study you've done. There's little point in us repeating material that's widely available in standard resources. If you've attempted to understand the subject and didn't understand something about what you read, telling us what your current understanding is and what difficulties you have in applying it here will help us give you better answers. $\endgroup$ – D.W. Aug 7 '17 at 5:31