Market Risk Early Warning Based on Deep Learning and Fruit Fly Optimization

Table 6

Prediction results.

Kernel function	Proportion of trainingsamples (%)	Proportion of testingsamples (%)	Optimal parameter value	G	F	AUC
RBF	90	10	c = 0.6509,  = 0 6666	0.6954	0.2023	0.6071
	80	20	c = 0.0836,  = 0.0806	0.7659	0.2224	0.7946
	70	30	c = 0.0908,  = 0.0798	0.7083	0.4576	0.7338
	60	40	c−0.7157, g−0.6259	0.7578	0.2888	0.7821
	50	50	c = 0.1384,  = 0.1355	0.7151	0.2143	0.7490
Polynomial	90	10	c = 2.1547,  = 0.7367, d = 6.8425, r = 0.9121	07071	0.2667	0.6843
	80	20	c = 3.0080,  = 0.8021, d = 2.0911, r = 1.1900	0.6157	0.2222	0.5429
	70	30	c = 4.5994,  = 0.4957, d = 0.9812, r = 1.8810	0.6279	0.4000	0.5514
	60	40	c = 2.6578,  = 0.7339, d = 1.8111, r = 2.7077	0.6684	0.0909	0.5786
	50	50	c = 0.2456,  = 0.0897, d = 0.0835, r = 0.0810	0.6673	0.2353	0.5677
Linear	90	10	c = 1.6133	0.7098	0.2349	0.7465
	80	20	c = 0.1	0.7641	0.2311	0.7935
	70	30	c = 0.0887	0.7761	0.4295	0.8024
	60	40	c = 15.6392	0.6488	0.2127	0.6530
	50	50	c = 0.1044	0.6954	0.2110	0.7108
Sigmoid	90	10	c = 5.7636,  = 0.7344, r = 0.7683	0.7011	0.2879	0.7243
	80	20	c = 1.9080,  = 0.6945, r = 0.8052	0.6970	0.1319	0.7154
	70	30	c = 3.3143,  = 0.5031 r = 0.8505	0.7061	0.3002	0.7212
	60	40	c = 2.4913,  = 0.7543 r = 1.1081	0.6267	0.2199	0.5436
	50	50	c = 0.2223,  = 0.0758, r = 0.0757	0.6443	0.1900	0.5981