Hi I am new to Pytorch.
In may training matrix, most of the features are continuous, but I have several symbolic features.
I want to use neural network.
Do I need to convert these features to the numbers or is there any better way to do that?
Thanks for your time and the help
by symbolic, you mean discrete features (instead of continuous). For all discrete features, make them into one-hot coded symbols, or you can use an
nn.Embedding and learn an embedding for each discrete feature.
Thank you very much for the help!
I will try one of the method you mentioned.