Skip to main content
. 2017 Aug 29;7:9900. doi: 10.1038/s41598-017-10324-y

Table 3.

The performance of Dong et al.’s model as well as the L1-regularized LR, random forest, and convolutional neural network models with active features after switching the datasets for training and testing.

Algorithm V-score for Switched Datasets
L1-regularized LR from Dong et al. 0.757
L1-regularized LR with active features 0.827
Random forest with active features 0.662
Convolutional neural network with active features 0.631