Experiments are created to encapsulate a data source. Creating an experiment allows you to manage and view many high level activities related to preparing your data for training. These include:
- Selecting features/target and datatypes
- Configuring your search space
- Viewing raw data and useful metrics about your data.
After you are finished with your selections you can run an Experiment Session. An Experiment Session is an instance of your Experiment based on your selections. The result of this is a leaderboard of pipelines.