在神经网络中获得恒定输出

时间:2017-02-20 09:11:52

标签: r neural-network backpropagation

我正在使用两个excel文件进​​行培训&测试目的。我在两个excel文件中都有与高度相对应的D值。 以下是我写的代码:

library(neuralnet)
library(readxl)

traininginput <- read_excel("Training_Data.xlsx")
net.D_value <- neuralnet(traininginput$D_value ~ traininginput$height,traininginput, hidden=3, threshold=0.01, rep = 100)


testdata <- read_excel("Test_data.xlsx")

net.test <- compute(net.D_value, testdata$height)

cleanoutput <- cbind(testdata$height,testdata$D_value,as.data.frame(net.test$net.result))

colnames(cleanoutput) <- c("height","Expected D_value","Neural Net D_value")

print(cleanoutput)

但是我得到测试数据所有高度的常量输出如下:

     height Expected D_value Neural Net D_value
1    366.21        7.9599610        4.622364013
2    531.28        0.8100586        4.622364013
3   1005.37        2.7099610        4.622364013
4   1138.91        1.0595700        4.622364013
5   1199.37        0.7001953        4.622364013
6    966.48       13.3095700        4.622364013
7    398.41       32.5500500        4.622364013
8    674.71        1.3598630        4.622364013
9    662.16       20.3398400        4.622364013
10   841.57        2.0302730        4.622364013
11  1205.85        1.3603520        4.622364013
12  1289.47        7.0498050        4.622364013
13   954.32        0.6601563        4.622364013
14   396.73       14.0900900        4.622364013
15  1173.71        0.5400391        4.622364013
16   933.28        2.6503910        4.622364013
17  1658.32       14.6503900        4.622364013
18  1172.39       27.1201200        4.622364013
19   967.09        2.6298830        4.622364013
20   597.37        0.2797852        4.622364013

任何人都可以指导我出错的地方并帮我写出正确的代码吗?

0 个答案:

没有答案