使用 tune.nnet 时预测功能的问题

如何解决使用 tune.nnet 时预测功能的问题

在尝试使用我的 tune 函数从 e1071 R 包预测的预测时遇到一些问题。 这是预测函数。训练数据是 Mnist 数据集。

NNmodel_tuned <- tune.nnet(training[,-1],training[,1],size = 3,MaxNWts = 100000,decay = 5e-4,tunecontrol=tune.control(cross=10))

tuned_results <- predict(NNmodel_tuned$best.model,data=y) 来自tuned_results 的输出是

22549 0.9999997
18559 0.9999997
8980  0.9999997
4399  0.9999997
32485 0.9999997
2959  0.9999997
2136  0.9999997
27818 0.9999997
9940  0.9999997
21650 0.9999997
27655 0.9999997
13800 0.9999997
29169 0.9999997
5456  0.9999997
2021  0.9999997
7432  0.9999997
11274 0.9999997
3527  0.9999997
9293  0.9999997
2173  0.9999997
28362 0.9999997
21587 0.9999997
6207  0.9999997
15786 0.9999997
18608 0.9999997
26198 0.9999997
5301  0.9999997
11918 0.9999997
40281 0.9999997
3493  0.9999997
11860 0.9999997
34454 0.9999997
8314  0.9999997
16123 0.9999997
10729 0.9999997
37223 0.9999997
22340 0.9999997
27772 0.9999997
14174 0.9999997```

This doesn't make sense to me.. Am I doing something wrong? 
the y variable is the labels from an unseen test set.