CODEWITHSUNDEEP
pythonmachine-learningclassificationh2o
sedioben
sedioben

Reputation: 945

How do I get the accuracy/precision of a h2o model?

I try to get the accurracy of my multiclass classifier using logistic regression.Is there any way to get the accuracy with a built-in function or do I have to write the function myself?

below my code so far:

multinomial_fit = H2OGeneralizedLinearEstimator(family="multinomial",max_iterations=100)

multinomial_fit.train(x=train_h2o_cro.columns[1:],y=train_h2o_cro.columns[0],training_frame=train_h2o)

prediction_glm_h2o = multinomial_fit.predict(test_h2o)

multinomial_fit.model_performance(test_h2o)

With the last line of code, I only get the mse and nothing else.

Thanks in advance.

Upvotes: 4

Views: 3280

Answers (2)

Lauren
Lauren

Reputation: 5778

Edited

Because accuracy is not currently implemented as Erin pointed out, your options for evaluating the performance of your model are limited to the functions that are available to the H2OMultinomialModelMetrics.

for example you could look at the .mean_per_class_error(), take a look at the multiclass confusion matrix model.confusion_matrix(data), or the log loss .logloss() to name a few.

Upvotes: -2

Erin LeDell
Erin LeDell

Reputation: 8819

This is currently unimplemented, but it makes sense to add this. Here is the JIRA ticket where you can track the progress.

Upvotes: 3

Related Questions