Golden Features have great predictive power.
They are created from existing features
by applying mathematical operators
The mljar AutoML framework
uses numeric features in Golden Features search.
From each pair of original features it creates a new feature by using of mathematical operators:
+, -, /, *.
The Decision Tree algorithm is used to assess the predictive power of newly created features.
Only top new features are included in the training data.
Golden Features procedure helps you discover unknown relationships in your data. Let yourself be surprised.
Improve your Machine Learning pipeline accuracy by including Golden Features in your dataset.