AutoML in Sparkflows

AutoML enables developers to train high-quality models for their use cases quickly. It automatically tries out a number of Algorithms and creates a leaderboard of the various algorithms tried.

Sparkflows provide support for H2O and PyCaret AutoML packages. Currently users can create the AutoML experiment by selecting the dataset, the ML-type (Regression, Classification etc), label column & engine type (H2O, Pycaret).

Configure Auto ML Experiment

Create the AutoML experiment by selecting the problem type, dataset, target column, model save path & package type.

Configure other parameters like features, algos etc

Train Models

Train the models by clicking on start experimenting,which will create the workflow and start running on the clusters

Model Result

When AutoML completes execution, the model details are displayed along with the leaderboard

View Parameters for each Model

Click on model_id to view the parameters of each model

Compare and Mark the best model

You can compare the different models and mark the best models for later use