在R中应用神经网络进行回归

Question

我在r中有以下示例数据框

a[::-1]

Count Year Hour Weekday day month week_of_month daycount w_y 23 2018 08 Wednesday 20 06 3 201 24 103 2018 12 Thursday 08 03 2 97 9 71 2018 09 Thursday 12 04 2 132 14 是我的目标变量，火车数据框的形状是Count，测试是c(932,9)

我正在使用以下代码在R中的keras中训练神经网络模型

c(310,9)

当我运行上面的代码时，它给我以下错误，

 model <- keras_model_sequential()

model %>%
  layer_dense(units = 8, input_shape = c(9)) %>% 
  layer_activation_leaky_relu() %>%
  layer_dropout(rate = 0.2) %>%
  layer_dense(units = 1) # output layer with linear activation

model %>% compile(optimizer = "adam", loss = "mean_squared_error")

### Data preparation 

dataSplit = function(data, p = 0.75) {
index = 1:nrow(data)
index = sample(index)
return(index[1:round(length(index) * p)])
}

index = dataSplit(train_df)
otrain = train_df[index, ]
otest = train_df[-index, ]

x_train <- otrain[,c(2:9)]
y_train <- otrain$Count

x_test <- otest[,c(2:9)]
y_test <- otest$Count

hist <-
  model %>% fit(
  x_train,
  y_train,
  batch_size = 10,
  epochs = 100
 )

我在哪里犯错？与 Error in py_get_attr_impl(x, name, silent) : AttributeError: 'list' object has no attribute 'dtype' 有关系吗？