- Create a new experiment.
- Select a data source for your business problem. It can be in the form of CSVs, time-series, etc. AutoML automatically infers the type of ML problem, classification, or regression.
- AutoML pre-processes the data and lists all the features with their respective statistics. Choose the features you wish to include in the training. Also, select the target class in case of a classification problem.
- Set training configurations, such as a number of trials, ML algorithms, values of hyperparameters, etc. Default values work just as well for your convenience.
- Hit “Run” to initiate training. AutoML trains various ML models, searching across a number of algorithms and hyperparameters and comes up with a summary of best performing models with their respective performance metrics. Choose the model of your choice to download and deploy as an endpoint.