Table 1.

The predictive results of different input parameters.

Algorithms	117 features	50 features	84 features	112 features
Naïve bayes	95.05%	94.75%	96.01%	94.85%
Logistic	97.12%	96.41%	96.77%	95.40%
IBK	89.39%	89.55%	91.06%	89.49%
J48	88.79%	89.55%	89.09%	88.43%
Random forest	95.61%	94.80%	94.64%	95.45%
libSVM	96.57%	96.57%	96.82%	96.87%

(In this table, the 50 features were selected by the mRMR, the 84 features with the P-values less than 0.001 were selected by ANOVA and the 112 features were selected by MRMD).