machine learning - which model is h2o.predict(aml@leader, test_df) using? -


after using automl generate aml leaderboard, ran

h2o.predict(aml@leader, test_df)  

but how can know model on leaderboard using? , if want access structure or hyperparameter of model on leaderboard how can so?

besides result on test set not 1 on validation set, common - did use wrongly or has tendency overfit?

also want understand infrastructure better, after h2o.init data transmit server in h2o.ai's clusters or happen on local laptop?

thanks.

it's using "leader" model, #1 model on leaderboard, ranked default metric ml task (binary classification, multiclass classification, regression). leader model id here: aml@leader@model_id.

the leader model, stored @ aml@leader, regular h2o model, if want @ parameters used, @ aml@leader@parameters parameters set, or aml@leader@allparameters parameter values (including ones did not set manually).

the validation_frame used tune individual models via stopping, validation error overly-optimistic compared test error, estimate of generalization error.

the third question out of scope post, i'll answer anyway. when using h2o , start cluster using h2o.init() running locally on laptop. if start h2o cluster somewhere else, such amazon ec2 or own remote servers, can pass ip address of server h2o.init() command using ip argument connect , computations run on remote machine. either way, servers entirely under control -- there no "h2o cloud" owned h2o.ai remote processing.


Comments

Popular posts from this blog

python Tkinter Capturing keyboard events save as one single string -

android - InAppBilling registering BroadcastReceiver in AndroidManifest -

javascript - Z-index in d3.js -