Scale-free dynamics via detrended fluctuation analysis (DFA)
Вставка
- Опубліковано 9 лип 2024
- This video lesson is part of a complete course on neuroscience time series analyses.
The full course includes
- over 47 hours of video instruction
- lots and lots of MATLAB exercises and problem sets
- access to a dedicated Q&A forum.
You can find out more here:
www.udemy.com/course/solved-c...
For more online courses about programming, data analysis, linear algebra, and statistics, see
sincxpress.com/ - Наука та технологія
This is great! Thank you so much for your explanation!
Very clear explanation. Thanks!!
Excellent video, thank you very much
Very nice presentation of the MFDFA. Really enjoyed it. It would have been amazing if you had displayed the 'follow-up' methods such as multifractal spectrum, too. Nonetheless, thumbs up for the video and your channel!
Thanks :)
I watched so many of your videos recently that i am pretty sure to hear sickness in your voice, and i started feeling sorry for you... 3 years later! 🤣But didn't change the quality of the explanation, thank you for sharing your work, this time frequency analysis playlist is gold. Thank you
Thank you for your concern, Florian :) I don't remember if I was sick when I recorded this video, but when I'm doing recordings, I like to power through them. Sometimes I also record videos early in the morning, so it could be my just-woke-up voice, lmao.
thank you very much , please could you prepare other course about multifractal analysis
very useful!
Useful lecture, thanks!
:)
Such an interesting analysis method. Thanks for making this video. I've become a big fan of your channel and have two of your books. Can you please recommend any articles on DFA that someone new to the topic can read?
There are many. I think this special issue would be a good place to start: www.frontiersin.org/research-topics/505/scale-free-dynamics-and-critical-phenomena-in-cortical-activity#articles
Thanks a lot for this wonderful video. I have one question- Is the hurst exponent same as DFA because in the last plot you mentioned slope to be Hurst exponent and the procedure of DFA analysis also considers the slope of linear fit as DFA value.
Yes, the exponential of the fit in log-log space is the estimate of the Hurst exponent.
the cummulative sum is not clearly specified - do you account for the size of each jump? Probably not - because then you would not have a downward trend a the beginning of the plot
I'm not sure which "jump" you are referring to. Anyway, the cumulative sum happens at each smoothing step, before segmenting.
@@mikexcohen1 at 4.31 you say that at the beginning there are more jumps going downwards - that made it a bit hard for me to understand. Now I see that the downward trend of the CUSUM is due to the fact that the majority of the data points in this segment are below the MEAN of the total window. So it is actually not that there are more jumps downward but rather the signals tendency to fluctuate below the MEAN value.
7:10 its confusing to see 8 windows but 24 graphs below. each graph is supposed to represent 1 window, right?
This has also confused me, could you clarify?
hi is there a code for DFA in python?
thanks
I would also be very interested in python code for DFA
Really interesting ! I am trying to compute DFA for my master thesis and I fail each time, may be you coul help me ??
I hope this video was helpful! Unfortunately, I don't have the time to do individual consultation or collaboration.
How the detrend at minute 7:00 is performed?
Hi Luca. Standard detrending is to fit an order-1 polynomial to the time series. I use the detrend() function in MATLAB.
@@mikexcohen1 ok thank you!
Excuse me I have another question: why DFA and Hurst exponent scale as an exponential?
It's common practice to both both axes in log-scale, and then do a linear fit. So, a linear fit in a log-log plot is actually an exponential fit.
Is there a python toolbox so I can build, plot, and learn?
Hmm, I'd guess there is a Python library for DFA, although I don't know off-hand. But the math isn't too difficult to implement on your own. It's definitely been implemented in MATLAB, so you could start from there and translate into python.