Each week, we select a recently published Open Access article to feature. This week’s article comes from Statistica Neerlandica and suggests The Dutch Draw baseline is optimal for binary classification.

The article’s abstract is given below, with the full article available to read here.

Pries, J., Bijl, E., Klein, J., Bhulai, S., & Mei, R. (2023). The optimal input-independent baseline for binary classification: The Dutch Draw. Statistica Neerlandica, 1– 12. https://doi.org/10.1111/stan.12297

Before any binary classification model is taken into practice, it is important to validate its performance on a proper test set. Without a frame of reference given by a baseline method, it is impossible to determine if a score is “good” or “bad.” The goal of this paper is to examine all baseline methods that are independent of feature values and determine which model is the “best” and why. By identifying which baseline models are optimal, a crucial selection decision in the evaluation process is simplified. We prove that the recently proposed Dutch Draw baseline is the best input-independent classifier (independent of feature values) for all order-invariant measures (independent of sequence order) assuming that the samples are randomly shuffled. This means that the Dutch Draw baseline is the optimal baseline under these intuitive requirements and should therefore be used in practice.