least squares method in the matlab statistics and machine learning toolbox  (MathWorks Inc)

Journal: G3: Genes | Genomes | Genetics

Article Title: Improved genomic prediction performance with ensembles of diverse models

doi: 10.1093/g3journal/jkaf048

Figure Lengend Snippet: A comparison of genomic prediction performance of the naïve ensemble-average (ensemble) model vs each of the individual genomic prediction models in violin plots. The width of the violins indicates the distribution of the metric values for predictions from all combinations of the 5 RIL populations, 3 training-test ratios, and 500 random samples. The performance of genomic prediction models was measured with a) the Pearson correlation and b) MSE. The orange represents the performance of classical models (rrBLUP, BayesB, and RKHS) while the green represents machine learning models (RF, SVR, and GAT). The red is the performance of the ensemble. Box plots within the violin plots represent the median metric value (white line) and the interquartile range (black box) with whiskers extending 1.5 times the interquartile range.

Article Snippet: From the various machine learning methods, we selected RF , SVR ( Drucker et al. 1996 ), and GAT ( Velickovic et al. 2017 ) for our investigation of ensemble prediction.

Techniques: Comparison