Application of machine learning algorithms and feature selection in rapeseed (Brassica napus L.) breeding for seed yield

Table 1 The performance of the algorithms to predict the SY of rapeseed using all measured traits

Algorithm	Kernel function /Loss function	Training			Testing
Algorithm	Kernel function /Loss function	R²	RMSE	MAE	R²	RMSE	MAE
Multiple Linear Regression (MLR)	–	0.856	0.247	0.191	0.786	0.329	0.254
Ridge Regression (RR)	–	0.843	0.258	0.198	0.830	0.294	0.234
Bayesian Ridge Regression (BRR)	–	0.846	0.255	0.196	0.825	0.298	0.236
Automatic Relevance Determination (ARD)	–	0.842	0.259	0.205	0.834	0.290	0.227
Generalized Linear Model (GLM)	–	0.849	0.253	0.194	0.809	0.311	0.243
Stochastic Gradient Descent (SGD)	Squared Error (SE)	0.809	0.285	0.222	0.839	0.286	0.224
	Huber	0.788	0.299	0.232	0.791	0.325	0.251
	Epsilon Insensitive (EI)	0.814	0.281	0.218	0.832	0.292	0.227
	Squared Epsilon Insensitive (SEI)	0.818	0.277	0.216	0.841	0.284	0.223
Nu-Support Vector Regression (NuSVR)	Linear	0.841	0.259	0.195	0.823	0.300	0.237
	Radial Basis Function (RBF)	0.847	0.255	0.194	0.841	0.284	0.219
	Sigmoid	0.813	0.282	0.213	0.809	0.312	0.246
	Quadratic Polynomial (QP)	0.861	0.243	0.194	0.860	0.266	0.210
	Cubic Polynomial (CP)	0.826	0.271	0.210	0.851	0.275	0.227
Epsilon Support Vector Regression (ESVR)	Linear	0.836	0.263	0.204	0.815	0.307	0.242
	Radial Basis Function (RBF)	0.819	0.277	0.211	0.841	0.284	0.223
	Sigmoid	0.685	0.366	0.273	0.738	0.356	0.259
	Quadratic Polynomial (QP)	0.848	0.253	0.193	0.846	0.279	0.220
	Cubic Polynomial (CP)	0.834	0.265	0.198	0.843	0.282	0.232
Linear Support Vector Regression (LSVR)	Epsilon insensitive (EI)	0.842	0.258	0.191	0.813	0.308	0.238
	Squared Epsilon Insensitive (SEI)	0.843	0.258	0.197	0.830	0.294	0.232

R² determination coefficient, RMSE root mean square error, MAE Mean absolute error

ISSN: 1746-4811