I'm curious as I'm going through the notebook, why did Ken or Jeff decide to do their split using almost a loc type variable compared to using train_test_split in sklearn? Is there some sort of time series component that I'm not aware of?

hey jonathan! sorry for the late reply here, but you can use sklearn version or use the .loc version. I use .loc, because then I can split based off the date of my choosing.

Follow this topic