There is a blueprint for a regression model I recently built. In this blueprint, the predictions from Elastic-Net Regressor (L2 / Gamma Deviance) are fed into Light Gradient Boosting on ElasticNet Predictions (Gamma Loss). Opening the documentation for the latter does not show how this is implemented. I would like to understand exactly how the predictions are used to build the Light GBM model. Please elaborate. Thank you.
The Elastic-Net Regressor (L2 / Gamma Deviance) feeds residuals from its predictions to the Light Gradient Boosting on ElasticNet Predictions (Gamma Loss) model. In this way, it is more like an added boosting step. More generally, all of our Elastic-Net nodes that feed into some boosting algorithm “on ElasticNet Predictions” pass the residuals of their predictions to the boosting algorithm.