Seagull: lasso, group lasso and sparse-group lasso regularization for linear regression models via proximal gradient descent

Table 1 Performance evaluation

R package	Accuracy parameter	R²	MSE	Non-zero	Time
SGL	10⁻⁴	0.91	12.78	8822	45 h 20 min
seagull	10⁻⁴	0.92	12.38	65,463	20 min
seagull	10⁻⁵	0.92	11.57	11,823	40 min
seagull	10⁻⁶	0.92	11.79	5095	2 h 13 min
seagull	10⁻⁸	0.92	11.84	5072	4 h 50 min

Accuracy parameter refers to a package-dependent convergence parameter; R² is the squared correlation coefficient and MSE is the mean squared error of chronological and predicted age; Non-zero denotes the number of CpG sites with non-zero effect estimate; Time is the computational time needed to calculate the full regularization path

ISSN: 1471-2105