3.3.4 RR , SGD , EN , KNN , Bagging Model
The Ridge Regression, Stochastic Gradient Descent, Bagging, K-Nearest Neighbor and Elastic Network models, which used the same feature selection method as Random Forest, all have a excellent result. The mean square error of RR, SGD, Bagging, KNN and ENM respectively were 0.028 , 0.070 , 0.039 , 0.070 , 0.028. The R-square respectively reached 0.91, 0.81, 0.92, 0.80, 0.93. The candidate activity prediction results obtained from the above five models were mostly superior to the control set. (Table 4)