Tutorial #2 general comments

voytekresearch / Tutorials

Learn to do common Voytek Lab things.

83 stars 21 forks source link

Tutorial #2 general comments #17

Open LeoVascelli opened 5 years ago

LeoVascelli commented 5 years ago

Filters are rad because...: Besides isolating rather narrow frequency bands filters can also restrict your signal to a broad frequency range. That is a super common use of filters.
Why not start with a tim-series signal that contains more than two oscillations? Way more realisitic and more eye-opening when taken apart into bands.
Maybe start with an acausal filter (filt instead of filtfilt) to later point out the shift caused by this
Impact of filter order! e.g. via Kernel width, make a note towards pre-stim vs poststim situations where smearing might be something you want to avoid / keep control of.
maybe point to papers: e.g. https://www.sciencedirect.com/science/article/pii/S0165027014002866?via%3Dihub
Not sure if "Complex_Signal" is a good name for that time series. Gets confusing when talking about complex numbers in a later tutorial (phase+amplitude measures)

rdgao commented 5 years ago

copied over from my (now closed) issue:

sine waves are incorrectly spelled as 'sin' in text
says 3 parameters for butterworth filter but 4 are listed. Lower and upper bounds can be combined into the same bullet.
time series plots are not the same size
kernels are incorrectly spelled as 'kernal'
arbitrary indexing magic numbers don't follow the (nice) convention of indices variables in other notebooks
gaussian window is incorrectly referred to as a 'gaussian distribution'
gaussian filter power needs to be normalized, otherwise filtered signal has a higher amplitude than the original
link to filtering tutorial in neurodsp at the end, which has more extensive coverage on frequency response of filters

also, agree with the above on not using "complex" and smearing. Smearing across trial boundaries, in particular, is not covered in ndsp tutorials and should be touched on

lakimenon commented 5 years ago

I think the paragraph that starts "In the plot below...." (right before the Butterworth filter parameters list) would be better placed under the plots so it's easier to look at them while reading. Note that it would then have to be changed to "In the plots above..."
As mentioned in another comment, I also noticed that kernel was misspelled as "kernal" and that the butterworth filter parameter section says there are three but lists four

meyhaa commented 5 years ago

Overall (up until content under construction)
- The format of this tutorial is different from the rest. Add a more detailed introduction with a list of the goals/ the purpose of this tutorial. (Reference tutorial 3)
- Most of the graphs/visualizations need to have their axes labeled.
- In general, sections of text need more subheadings/bolded terms or phrases.

Potential Key Terms for Glossary (list for you to narrow down/edit): Filter Amplitude Frequency Frequency band (lower/upper bounds) Complex wave Butterworth filter Order (of Filter)