0
$\begingroup$

I have dataset of gold prices and after modifying and some preprocessing i ended up with dataframe below: enter image description here

There is 50,000 record in dataset and all columns expect date are int type and date is datetime object. i need to predict price per unit in some specific dates. but somehow i baffled with so many methods.

My question is what data mining algorithm/method is results good prediction for this kind of data ?

$\endgroup$
  • $\begingroup$ It is a time series with covariates. I would try first an ARIMAX model $\endgroup$ – MasterJedi Nov 23 '16 at 3:48
  • $\begingroup$ 50k and only a few columns sounds small enough to be loaded and played with in Orange. $\endgroup$ – K3---rnc Nov 23 '16 at 23:52
0
$\begingroup$

Recurrent Neural Network (RNN) are quite efficient at dealing with sequential information such as time-series.
LSTM (Long Short Term Memory) is a type of RNN that can learn long-term dependencies and overcome some problems that vanilla RNNs have.
If you want to try it on your problem, I'd suggest this tutorial on LSTM for time-series prediction.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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