tensorflow2.0 - tensorflow dataset string/bool transformation for both bool & string values how to

Question

i have a tensorflow dataset that contains several elemets:

string, int & bool

for element in train_df.take(1):
   print(element)

(OrderedDict([('textfield', <tf.Tensor: shape=(1,), dtype=string, numpy=array([b'240300'], dtype=object)>), ('boolfield', <tf.Tensor: shape=(1,), dtype=bool, numpy=array([ True])>), ('intfield', <tf.Tensor: shape=(1,), dtype=int64, numpy=array([1853])>)]), <tf.Tensor: shape=(1,), dtype=int64, numpy=array([1])>)

i wonder how i can convert them before training to numeric values.

• lets say the string has the values "True" or "False" so should be 1 or 0

• and same logical for the bool field

i've saw that are several techniques like:

• layer transformation
• dataset transformation
• tfx

but im not sure how to do it and when to use each of them

can you assist please?

question from:https://stackoverflow.com/questions/65831901/tensorflow-dataset-string-bool-transformation-for-both-bool-string-values-how

Answer 1

Waitting for answers