Time Series Analysis

[quantmenot]

I am looking to do analysis on a time series.

Can somebody compare ARIMA,SVRs, XGBoost, and LSTMs for the task?

Has anyone ever tried ensemble all of these under an AdaBoost?

 

[MikeLawrence]

I only know one of these techniques, but I'm pretty sure that your question is far too general for us to be of use.

 

[quantmenot]

I only know one of these techniques, but I'm pretty sure that your question is far too general for us to be of use.

The question is adequate. If you have experience with a specific technique, you should be able to speak to it's strengths and weaknesses for a particular time series task -- regression, reconstruction, denoising, decomposition. If you have experience with multiple techniques, you should be able to compare their strengths and weaknesses.

 

[_quanty_]

The question is adequate. If you have experience with a specific technique, you should be able to speak to it's strengths and weaknesses for a particular time series task -- regression, reconstruction, denoising, decomposition. If you have experience with multiple techniques, you should be able to compare their strengths and weaknesses.

I think Mike is pointing out that you should go ahead and do your own research on this extremely broad question instead of relying on someone else to spend hours researching it for you. Entire college courses are taught on these techniques, someone cannot summarize all of their strengths and weaknesses for you in a blog post.

 

[quantmenot]

I think Mike is pointing out that you should go ahead and do your own research on this extremely broad question instead of relying on someone else to spend hours researching it for you. Entire college courses are taught on these techniques, someone cannot summarize all of their strengths and weaknesses for you in a blog post.

OK, feel free not to answer.

 

[MikeLawrence]

The question is adequate. If you have experience with a specific technique, you should be able to speak to it's strengths and weaknesses for a particular time series task -- regression, reconstruction, denoising, decomposition. If you have experience with multiple techniques, you should be able to compare their strengths and weaknesses.

Fair, but I'm not the one to do it. And @_quanty_ is right only in that it might be a minute before someone comes along who is capable and willing to give a detailed answer to this. I hope you find some source of help though.