Building an Lstm model and reshaping data

Question

Is scaling my data necessary

How can I make the input shape for my lstm model take 3dim and how do I get rid of the None value?

For my input shape I am getting (?,100,100) I want it to be (473,100,100)

After importing my data X,Y.

I split it into 3d shape to represent X=(n_samples,timestamps,features). Y remains unchanged.

Then I split and train my data

Then I scaled my data.

When modelling an lstm model i can only input the shape of (timestamp, feature) if i try make it (n_samples,timestamps,features) i get an error saying i gave 4dim instead of 3dim. With the shape (None, 473,100,100)