The primary purpose of machine learning algorithms is to define a function
which predicts an unknown phenotype based on a genotype observation based on sample data

A major reason for the adoption of machine learning algorithms is that they are well suited for developing predictive models when the number of features is larger than the number of samples.

A standard characteristic of GWAS results is that the number of attributes (p) greatly outnumber the number of sample points (n). This is usually described as the curse of dimensionality or the large p and small n problem.Ideally, it is a problem for classical multivariate regression.