Conditional Super Learner

TL;DR

The paper introduces the Conditional Super Learner (CSL), an algorithm that adaptively selects the best model based on covariates, combining cross-validation and meta learning, with proven convergence and strong empirical performance.

Contribution

It presents a novel CSL algorithm that finds a local minimum efficiently, with theoretical convergence guarantees and empirical evidence of its effectiveness over traditional stacking methods.

Findings

Converges faster than $O_p(n^{-1/4})$.

Outperforms stacking in empirical tests.

Suitable for hierarchical modeling.

Abstract

In this article we consider the Conditional Super Learner (CSL), an algorithm which selects the best model candidate from a library conditional on the covariates. The CSL expands the idea of using cross-validation to select the best model and merges it with meta learning. Here we propose a specific algorithm that finds a local minimum to the problem posed, proof that it converges at a rate faster than $O_p(n^{-1/4})$ and offers extensive empirical evidence that it is an excellent candidate to substitute stacking or for the analysis of Hierarchical problems.