Lab - Model Creation/AutoML

The next step in the process is to use Kortical's AutoML to create candidate models from the datasets you've prepared during exploratory data analysis. Model creation involves:

• Preprocessing and cleaning input data
• Creating features that are appropriate for machine learning
• Training potential models with cross-validation
• Evaluating those models across a range of metrics

Each step is highly configurable through Kortical's language feature, which allows data scientists to control the training process as much or as little as they need.

Kortical's AutoML automatically optimises across the very large space of possible feature engineering choices, model types and model parameters to find the best choices for the problem at hand. The trained models can then be explored and published on the model candidate leaderboard. Once published, a model is ready for predictions to be made on new data and its behaviour can be explained to understand its key drivers.

Read about the Lab page to get started with model creation.