I am using H2O models for both training and testing and while I can view the model's performance on the training dataset, I also want to assess its performance on the test dataset.
Hello Chris, You can achieve this by following these steps:
When splitting the data into training and test sets, make sure to save the test data along with the trained model.
In the subsequent steps, use the saved trained model to make predictions on the test data and save these predictions.
Create a PySpark workflow to read the predicted output and apply custom evaluation metrics to select the desired evaluation metrics.