调参代码_np.arange(0.01, 0.1, 0.01)

#参数的值
        test_value=[np.arange(0.001,0.01,0.001),np.arange(0.01, 0.1, 0.01),np.arange(0.1, 1, 0.1),np.arange(1, 10,1),np.arange(10, 100, 10),np.arange(100, 1000, 100)]
        for j in np.arange(0, len(test_value), 1):
            for k in test_value[j]:
                result = []
                prior = {
                    'zeta':k,#测试zeta这个参数
                    'u':0.07, 
                    'v':0.5, 
                    'gamma': 300,
                    'max_iter' :100,
                    'mix_threshold':0.1,
                    'v_test':0.06, 
                    'model_zeta':0.004,  
                    'model_u': 60,  
                    'model_v':8,  
                }
                args.prior = prior

                result = []
                #每个参数跑五次，结果存入result
                for n in range(0,5):
                    trainer = Trainer(args)
                    pred, _ = trainer.train(train_x[np.newaxis,:,:])
                    print(set(pred))
                    acc = cluster_acc(train_y, pred)[0]
                    print("acc",acc)
                    result.append(acc)
				#求五轮的平均值
                average=np.average(np.array(result))
                print("--------------------------------------average:",average)
                # 将平均值到max中
                max.append(np.array([average,k]))
        # 找到max中最大
        t1=np.max(np.array(max)[:,0])
        max_index=np.where(np.array(max)[:,0]==t1)[0]
        print("max_result:",t1,np.array(max)[max_index,:])
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37