¶ Processing the data in Python
scipy.signal.find_peaks_cwt
finds all local maxima (possibly the peak of an SCR) in each time series
we can then filter out any peaks which did not occur within 2 seconds of another person's
the remaining peaks are possible "meaningful moments"
sklearn.neighbors.kde.KernelDensity
finds clusters of these possible synchronizations
then moviepy to cut into the correct video clips
¶ AcqKnowledge
there is a lab with some legit software on Cornell's campus (256 HumanEcol)
¶ Algorithms
- advanced deconvolution method estimates times of stimuli, magnitude, and physiological characteristics even
- I would likely have to ask them for the code, and try to modify it
- sounds extremely complicated
- simpler deconvolution method probably doesn't work for me, because I don't know when the stimuli are
- surrogate data generation?? I'm dubious...
- why cross-correlation isn't good enough
- a non-cvxEDA method for extracting tonic and phasic
¶ Electrodermal activity in python
- EDA explorer looks great, but only supports Q Sensor, E4, and Shimmer data. Need to reverse engineer these formats to use it...
- cvxEDA looks very well done, and exactly what I need
- NeuroKit has analysis software for multiple types of neuro data, and uses cvxEDA under the hood for EDA (EDA docs)
- I can at least get the data into NeuroKit
- Stimfit (software) (download)
- this looks good, but can't import CSVs!
- this looks good, but can't import CSVs!
- maybe SigViewer?
- General Data Format has no implementation in Python, but it does in MatLab
- MNE looks fabulous!!
- also, I should find my old notes (lol)
- FOUND THEM! but in a text file... here you go
- HDF5 is apparently "non-proprietary" -- so there's a converter
¶ New notes on software for SCR analysis
¶ Old notes on software for neuro in general
- Software for neuroscience computing
- Where I found neuroDebian
- The Ultimate Neuroscience Software Platform
- GREAT!
- Neo - NeuralEnsemble
- Electrophysiology analysis package
- Byron Yu: Software
- Amazing data visualization and analysis strategies for neural data
- DataHigh
- One in particular I liked ^^^^^^^^
- ConnectomeDB
- HUGE database of neural stuff.
- THIS!
- NITRC Computational Environment
- NITRC-CE is a virtual computing platform pre-configured with many neuroimaging data analysis applications.
- I tried this out and it's STELLAR.
- The VirtualMachine is just run IN BROWSER
- automatically connected to tons of nice data.
- needed to put "archive" in the sources.list -- look it up, it's just an old version of Debian