Skip to main content
. 2020 May 21;37(14):2012–2016. doi: 10.1093/bioinformatics/btaa535

Table 3.

Cross-validated logistic deviance for binary classification problems (rows) under different regularization methods (columns), with the class frequencies (0/1) in the first two columns, and the results for post-hoc feature selection in parentheses

#0 #1 Ridge Lasso Tune Stack
BLCA 19 389 0.242 0.248 0.265 0.240 (0.240)
BRCA 112 977 0.271 0.197 0.187 0.266 (0.276)
CHOL 9 27 0.754 0.069 0.061 0.023 (0.645)
ESCA 11 172 0.394 0.296 0.302 0.259 (0.260)
HNSC 44 475 0.279 0.291 0.264 0.267 (0.269)
KICH 25 41 0.853 0.755 0.608 0.581 (0.584)
KIRC 72 460 0.198 0.155 0.162 0.159 (0.159)
KIRP 32 257 0.390 0.250 0.299 0.255 (0.252)
LIHC 50 321 0.362 0.330 0.304 0.323 (0.330)
LUAD 59 457 0.262 0.245 0.253 0.206 (0.203)
LUSC 51 450 0.161 0.076 0.078 0.073 (0.073)
PRAD 52 444 0.444 0.438 0.464 0.442 (0.488)
STAD 35 383 0.339 0.237 0.230 0.237 (0.237)
THCA 59 434 0.431 0.393 0.463 0.458 (0.459)
UCEC 10 360 0.088 0.072 0.085 0.062 (0.059)