A Genetic Algorithm Based Support Vector Machine Model for Blood-Brain Barrier Penetration Prediction

Table 1

Performance comparison of models with different number of features.

Number of features	Training (CV = 10)		Prediction/		Parameters of SVM
	MSE		Test set	Training set
4	0.1197	0.674	0.722	0.740	38.8833	0.6081	0.1491
5	0.1042	0.715	0.770	0.805	16.3419	0.7973	0.2743
6	0.0945	0.744	0.840	0.829	13.3573	0.7158	0.1513
7	0.0959	0.74	0.821	0.843	34.3067	0.5218	0.1595
8	0.0883	0.761	0.834	0.883	60.9596	0.5871	0.2357
9	0.0815	0.777	0.847	0.864	3.7770	0.8764	0.1663
10	0.0823	0.776	0.858	0.903	15.2236	0.6247	0.1434
11	0.0714	0.804	0.861	0.891	5.6937	0.6531	0.1573
12	0.0780	0.787	0.864	0.905	7.2787	0.7428	0.1515
13	0.0817	0.778	0.862	0.922	4.1957	0.7791	0.1574
14	0.0812	0.778	0.882	0.917	14.8391	0.5002	0.2054
15	0.0734	0.799	0.870	0.919	4.9915	0.5231	0.1077