Ensemble Methods with Voting Protocols Exhibit Superior Performance for Predicting Cancer Clinical Endpoints and Providing More Complete Coverage of Disease-Related Genes
Table 4
Voting results of KIRC.
Platforms
Class
Vote from all methods
Vote after filtering
MCC avg
MCC max
KIRC-mRNA
bayes
0.4216
0.3672
0.2401
0.6590
functions
0.6292
0.7162
0.0399
0.7785
lazy
0.4682
0.6292
0.3801
0.8474
meta
0.5261
0.5261
0.3780
0.7785
misc
0.4176
0.4105
0.2991
0.6058
rules
0.4385
0.6110
0.0723
0.6885
trees
0.5421
0.5421
0.3649
0.7785
overall
0.6110
0.5421
0.2535
0.8474
KIRC-miRNA
bayes
0.2368
0.1805
−0.0433
0.5131
functions
0.0889
0.0530
−0.1698
0.7162
lazy
0.4371
0.3410
0.1865
0.5249
meta
0.1667
0.1667
0.1105
0.4542
misc
0.4606
0.4795
0.3230
0.6590
rules
0.0889
0.2689
−0.0795
0.4606
trees
0.0530
0.1667
0.1165
0.6110
overall
0.1667
0.1667
0.0323
0.7162
KIRC-mRNA and KIRC-miRNA
bayes
0.4795
0.4795
0.0984
0.6590
functions
0.2605
0.6885
−0.0649
0.7785
lazy
0.5514
0.4371
0.2833
0.8474
meta
0.2300
0.2300
0.2442
0.7785
misc
0.4105
0.5131
0.3110
0.6590
rules
0.2605
0.5249
−0.0036
0.6885
trees
0.4371
0.5249
0.2407
0.7785
overall
0.4371
0.5249
0.1429
0.8474
All of the measurements in the tables are MCCs, and the vote after filtering is the MCC based on the eliminated methods. The “avg” is the average of the MCCs.