0
$\begingroup$

I would like to learn time series data analysis and forecasting. I am knowledgeable in machine learning and have a good knowledge of deep learning (including RNN's, LSTM). I came to know that time series is a lot different from normal data and hence most of the preprocessing, feature extraction methods cannot be used for time series data and also that the working of time series algorithm is a little different than the normal machine learning algorithms.

Even though I read a lot of articles was still left confused about time series analysis and forecasting (like in cases of removing making data stationery I couldn't fully understand why and if it meant only for some simple time series algorithms).

Where can I learn time series forecasting (preprocessing, feature extracting, and ml/dl for time series data).

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

Actually, time-series forecasting is not trivial. It takes time to wrap your head around when to use what kind of model. Hyndman's online book is a good place to start. If you are looking for a simple one-page summary or roadmap that will give you a set of cookbook recipe's of when to use what, you will be looking for a while.

Improve this answer
$\endgroup$
4
  • $\begingroup$ Thanks. I think one page summary won't work in the whole area of machine learning especially when working with real world datsets. $\endgroup$
    – insomniac
    Commented Jan 31, 2022 at 9:53
  • $\begingroup$ I found the sktime package documentation and tutorials useful. The following two talks were pretty useful: youtube.com/watch?v=_vQ0W_qXMxk&t=731s , youtube.com/watch?v=v5ijNXvlC5A&t=1308s These are more statistical approaches rather than deep learning-based approaches. $\endgroup$
    – Rajiv Sambasivan
    Commented Feb 1, 2022 at 5:34
  • $\begingroup$ I've gone through the sktime documentation but found most of the tutorials theyve provided are based on univariate data and multivariate classification. i couldnt find much on multivariate forcasting. thanks for the vidoes $\endgroup$
    – insomniac
    Commented Feb 1, 2022 at 9:57
  • $\begingroup$ Not sure if the multiple terminologies are the cause of confusion. Time series that use exogeneous variables are supported, see sktime.org/en/stable/examples/… See the note about the work-flow, you just have an X argument in addition to the Y argument. $\endgroup$
    – Rajiv Sambasivan
    Commented Feb 1, 2022 at 11:46

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.