What are the best options for a big dataset classification? I am thinking of 2 solutions-


2.PCA(Principal Component Analysis)

I think the first approach is better.Does it works for my problem?Or maybe someother solution is better?

Thank you

  • $\begingroup$ Just use a Convolutional Neural Network. $\endgroup$
    – Gerry P
    Mar 28, 2020 at 17:53
  • $\begingroup$ @Gerry P can you give me a sample of what you mean?Thank you $\endgroup$
    – scoute21
    Mar 28, 2020 at 21:03

1 Answer 1


None of the methods you described may classify a dataset alone whereas both can be used to transform your data into another domain in an unsupervised fashion.

PCA projects your data onto n-orthogonal components. A trained encoder (first component of the autoencoder) can project your data onto a latent space.

Both of those representations can be used in conjuction with a classifier, such as a decision tree, to form a classification pipeline.

There is no globally best solution, try them both for your specific problem. PCA is, in my opinion, way faster to try out using sklearn instead of building your own autoencoder using PyTorch or Keras. Give PCA a first go and if the results are poor, try using an autoencoder.

  • $\begingroup$ @picko1 thank you for your answer i would like to see if someone has something else to say $\endgroup$
    – scoute21
    Mar 28, 2020 at 21:05
  • 1
    $\begingroup$ This is a great answer. While the autoencoder can give you a better result in the end, it is not fore sure and takes more time. In my experience performance is often similar, but it is very dependent on the domain. Be aware that PCA is not robust against outliers though. Training an autoencoder is also quick so you should definetly compare them :) $\endgroup$ May 10, 2020 at 11:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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