-2
$\begingroup$

This may sound like a stupid question, but trust me it's not, I am searching for a ml framework/library that makes you do most of the work yourself so I can learn rather than depending on a framework/library but STILL I do not wanna code ALL ml algorithms from scratch as that would be pretty painful

Hope that clears everything up, thanks in advance.

$\endgroup$
  • $\begingroup$ There are a lot of introductory options out there. Something like sklearn might interest you. But your question doesn't really make sense as all libraries aim to help you not restarting from scratches. $\endgroup$ – lcrmorin Oct 17 at 19:09
  • $\begingroup$ @icrmorin what do you mean by "introductory" , do you mean it cannot be used for making real projects? $\endgroup$ – AmirWG Oct 17 at 20:16
  • $\begingroup$ It is more than enough to build a real project. It is simpler than other framework that are specialised on one topic, that usually require custom installation / custom hardware and, more importantly for you, have smaller communities that can help you with obscure bugs. $\endgroup$ – lcrmorin Oct 17 at 20:20
0
$\begingroup$

Sklearn allows you to build your own pipeline steps (and estimators if you so choose). If you get comfortable with how the Pipeline object framework works it's infinitely configurable. It also allows you to rely on pre-built algos. To operate this way you will need to be comfortable with object-oriented programming in python rather than a script-based approach. Take a look at how to create custom pipeline steps in this helpful blogpost: https://towardsdatascience.com/pipelines-custom-transformers-in-scikit-learn-the-step-by-step-guide-with-python-code-4a7d9b068156

| improve this answer | |
$\endgroup$
  • $\begingroup$ may i know if it's relevant/popular? and sure i am comfortable with object-oriented programming , it's not me first time programming. $\endgroup$ – AmirWG Oct 17 at 20:17
  • $\begingroup$ I'd say it's one of, if not the most, popular machine-learning frameworks out there. The one exception is in deep learning which is typically done with PyTorch or Tensorflow $\endgroup$ – Oliver Foster Oct 17 at 20:51
  • $\begingroup$ so you are saying you can use sklearn for normal machine learning but tensorflow for deep learning? $\endgroup$ – AmirWG Oct 17 at 20:55
  • $\begingroup$ You can use sklearn for both but sklearn is limited in it's deep learning capabilities (it has estimators that cover some basic deep-learning architectures but if you want to get further into deep-learning design you will need a different framework like PyTorch or Tensorflow) $\endgroup$ – Oliver Foster Oct 17 at 20:58

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