Reshaping multivariate supervised data for LSTM in Keras

Question

I am having a hard time wrapping my head around reshaping my binary classified data to input to a Keras LSTM. Thought process:

Example DataFrame: