1

Is it possible to use data stored in Sframe to train e.g., a Random Forest, of scikit-learn implementation without converting the whole dataset to numpy?

user3689757
  • 57
  • 6
  • No. I also think it is even impossible with the more equal SArray. sklearn/numpy is using a lot of assumptions about the data, so if your data-format just differ in one aspect it is not going to work. – sascha May 13 '16 at 21:49

1 Answers1

1

According by Turi-forum:

"If you use the most recent version of SFrame (which only became available via pip yesterday) you can use the tonumpy function to create an ndarray from an SFrame."

https://forum.turi.com/discussion/1642/

Miklós Molnár
  • 764
  • 6
  • 5
  • It is seems this solution needs the Graphlab create library, which is not longer available, I wonder how would be the solution to use only SFrame open source. – Germán Alfaro Mar 22 '17 at 09:05