Could you please provide an example for general workflow?

[jacktang]

Hello there!

Thanks for the General Workflow and tips in examples dir, and could you please provide an example for general workflow?

[manu-mannattil]

By an example if you mean an example involving a specific real-world data set, I think that would not be very useful. Nonlinear time series analysis isn't a set of fixed methods and very often you would have to tailor your analysis techniques depending on your data set. Before using the module, I would suggest understanding the basic theoretical techniques involved (perhaps by reading a review paper or a book). Without understanding basic concepts, using the module would be counterproductive. Once you understand the techniques, then there are many examples in the examples directory that should help you use the functions in the module. Hope this helps.

[jacktang]

Hello @manu-mannattil , thanks for you kindly suggestions! I meant an example involved the whole worflow. In the example there is an assumption that the dataset need to be applied such step, but not answered the question: why need this step?/what's problem if this step is not involved? As you suggested, maybe some paper or book will answer my question :), could you please recommend some reading materials? Thank you!

[manu-mannattil]

I'm not sure I fully understand your question. Nonlinear time series analysis was historically developed to ascertain the presence of low-dimensional chaos and/or determinism in experimental data. However, the exact methods you would use would depend on your goals and the data you want to analyze. A good introduction to the techniques involved is the book by Kantz and Schreiber ["Nonlinear Time Series Analysis" (Cambridge, 2003)] and references in that book.

The "general workflow" I outlined in the examples section is just a guideline, based on my experiences when I wrote two papers that made use of these techniques. There is no requirement that you follow it.

Perhaps I would be able to better understand your question if you can tell me why you want to use nonlinear time series analysis. Is there a specific data set you want to analyze? Do you suspect the presence of chaos/determinism? What kind of indicators of determinism are you interested it?

[jacktang]

A good introduction to the techniques involved is the book by Kantz and Schreiber ["Nonlinear Time Series Analysis" (Cambridge, 2003)] and references in that book.

Thanks for the book!

Is there a specific data set you want to analyze?

Yes, I try to apply some nonlinear time series analysis ideas on Hourly Energy Consumption public dataset.

[manu-mannattil]

OK. I would still recommend familiarizing with basic concepts before using NoLiTSA. Part of the reason I say this is because people have used nonlinear time series analysis in the past to make outlandish claims, e.g., in the analysis of medical signals (EEG, ECG, etc.). Unfortunately, nonlinear time series analysis has only been modestly successful in extracting useful conclusions from most data, and it's easy to make mistakes if one is not careful enough.