Having problem with tensorflow 2.0.0 : Unimplemented: Cast string to float is not supported

[SomyKamble]

history <- model %>% fit(trainx,y_train,epochs=50,batch_size=5,validation_split=0.2)
2019-11-13 18:03:25.393757: W tensorflow/core/framework/op_kernel.cc:1599] OP_REQUIRES failed at cast_op.cc:123 : Unimplemented: Cast string to float is not supported
2019-11-13 18:03:25.396018: E tensorflow/core/common_runtime/executor.cc:642] Executor failed to create kernel. Unimplemented: Cast string to float is not supported
[[{{node Cast}}]]

reticulate::py_config()
python: C:\conda\envs\r-reticulate\python.exe
libpython: C:/conda/envs/r-reticulate/python36.dll
pythonhome: C:\conda\envs\R-RETI1
version: 3.6.9 |Anaconda, Inc.| (default, Jul 30 2019, 14:00:49) [MSC v.1915 64 bit (AMD64)]
Architecture: 64bit
numpy: C:\conda\envs\R-RETI1\lib\site-packages\numpy
numpy_version: 1.17.4
tensorflow: C:\conda\envs\R-RETI~1\lib\site-packages\tensorflow_init_.p

python versions found:
C:\conda\envs\r-reticulate\python.exe
C:\Users\somy\AppData\Local\Programs\Python\Python38\python.exe
C:\conda\python.exe
C:\Users\somy\AppData\Local\Programs\Python\Python38\python.exe

tensorflow::tf_config()
TensorFlow v2.0.0 (C:\conda\envs\R-RETI~1\lib\site-packages\tensorflow_init_.p)
Python v3.6 (C:\conda\envs\r-reticulate\python.exe)

[dfalbel]

This looks like a bug in your code. what's trainx? Is it a numeric matrix?

[SomyKamble]

Hi @dfalbel ,

Thank you for your reply .
I haved solved the issue, I guess you were right there is a bug in my code :
I request you to please refer to the below R code and share your valuable thoughts for the same.

library(mlbench)
library(keras)

data2<-data(BreastCancer)
data2<-BreastCancer
data2<-na.omit(data2)
data2<-data2[,-1]

for(i in 1:ncol(data2)){
data2[,i]<-as.numeric(as.factor(data2[,i]))
}

data2<-as.matrix(data2)

dim(data2)
data2[,10]<-as.numeric(as.factor(data2[,10]))

split<-sample(1:2,nrow(data2),prob = c(0.8,0.2),replace = T)

train<-data2[split==1,]

test<-data2[split==2,]

trainx<-train[,1:9]

train_y<-as.numeric(as.factor(train[,10]))

train_y<-train_y-1

y_train<- to_categorical(train_y)

testx<-test[,1:9]
test_y<-as.numeric(as.factor(test[,10]))
test_y<-test_y-1
y_test<-to_categorical(test_y)

model<-keras_model_sequential()

model %>%
layer_dense(units =512, activation = 'relu',input_shape = 9) %>%
layer_dense(units = 10, activation = 'relu')%>%
layer_dense(units = 10, activation = 'relu')%>%
layer_dense(units = 2, activation = 'sigmoid')

model %>% compile(
loss='binary_crossentropy',
optimizer='adam',
metrics=c('accuracy')
)

summary(model)

trainx<-as.matrix(trainx)
class(trainx)
history <- model %>% fit(trainx,y_train,epochs=50,batch_size=5,validation_split=0.2)

plot(history)

score <- model %>% evaluate(testx, y_test)

cat('Test loss:', score$loss, "\n")
cat('Test accuracy:', score$acc, "\n")

sam<-predict_classes(model,testx)

library(caret)

table(sam,test_y)
confusionMatrix(table(sam,test_y))