1
$\begingroup$

https://towardsdatascience.com/speed-testing-pandas-vs-numpy-ffbf80070ee7 (You can open the link in incognito if its locked).

Numpy arrays are faster than DataFrame on normal mathematical operations.

Should I use np arrays to train my algorithm? Or go for DataFrame? I understand DataFrame makes it easier to 'look' at the data. But will np array help in training?

$\endgroup$
1
  • $\begingroup$ What are you training, with what package? $\endgroup$ Sep 26 at 14:10
0
$\begingroup$

For TensorFlow, you need numpy arrays, or tensors as input. Here is the documentation for it and there are bunch of options when it comes to arguments for the fit method and it has to be an array, tensor at the most basic level or some generator that returns an object of a similar type.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.