I would like to plot a loss vs epoch graph from a deep quantile regression model in H2O. I’m using the H2ODeepLearningEstimator but can’t seem to find a way to retrieve the loss like in Keras.
https://machinelearningmastery.com/display-deep-learning-model-training-history-in-keras/
Could somebody please point me in the right direction?
Advertisement
Answer
I believe you are looking for the learning curve plot. In h2o you can retrieve the “scoring history” data using model.scoring_history()
.
You can influence the frequency of scoring as well as amount of data used for the loss estimate using parameters prefixed with score_
, e.g., score_interval
, score_validation_samples
etc.