Does Einstein Discovery require all potential collinear fields to be removed before executing the build story?

Einstein Discovery is designed to handle collinearity in data, which refers to the situation where two or more predictor variables are highly correlated. In this context, while it is generally ideal to eliminate collinearity prior to building a model to enhance interpretability and ensure stability, Einstein Discovery has built-in mechanisms to deal with this scenario effectively.

When collinear fields are present, Einstein Discovery will not fail outright during the model-building process. Instead, it provides a warning after the model is built if collinearity is detected. This notification alerts the user to potential issues without preventing the build altogether. Furthermore, the use of ridge regression within one of its algorithms helps mitigate the impact of collinearity by applying a penalty to the size of the coefficients, allowing the model to manage correlated predictors more robustly. Hence, while addressing collinearity is a best practice, it does not render the model inoperable.

This understanding emphasizes how Einstein Discovery is formulated to accommodate real-world data scenarios, providing flexibility and resilience to data intricacies like collinearity.