Nov 7th.

[rmarren1]

@rmarren1:

• [x] - in the webservice, include pre-processing steps and a (separate) pre-processing/quality control report
• [x] - work with @nkumarcc to integrade KDE bad electrode detection into webservice
• [x] - various small changes before demo
  • [x] -separate eeg-related functionality from the 'analysis' rest call, so that later it will be easier to separate timeseries stuff from eeg stuff.
  • [x] - various chart edits for post-processing in jovos coment in #18
  • [x] -make analysis links persist (set up a database, or just search the 'results' folder and auto-populate the 'view' table on user load)
  • [x] - refactor code to make it easier to add things
• [x] - randomized PCA code.md (or really anything that doesn't explicitly factorize the matrix. i think finding the top k eigenvectors with the power iteration method, where k is how ever many eigenvectors we need for the cumulative variance explained to be nearly 1, which was around k=10 for the EEG data)

@nkumarcc

• [x] - integrate KDE bad electrode detection into webservice - DoD is working output of bad electrode data in demo
  • [X] - better understand cross validation (which is being used for bandwidth selection in KDE). Need to see if cross validation takes too much time on real data - .md describing cross validation and description of times taken on real data
• [X] - Finish coherence code.md, with correct quantitative metric to see how well it works - .ipynb
• [ ] - Reach: Look into kurtosis and spectra decision making for bad electrodes - LaTeX of concepts and pseudocode
• [ ] - Reach: quantitative metrics for distributions to evaluate KDE better than simply qualitatively- LaTeX + potential implementation in .ipynb

@tsunwong123

• [x] - Pseudo-code for Robust PCA due 11/02, determine what to do from there. Some necessary information .md and LaTeX
• [x] - code.md for Robust PCA
• [ ] - Reach: Implement the Robust PCA Algorithm as part of the pipeline.

@mnatenz1

• [x] ICA working, with a way to visually analyze the independent components, and select which to remove from the original signal

[tsunwong123]

@gkiar

[gkiar]

👌

[jovo]

• [ ] ICA can subsample
• [ ] do we have design specs for how long things should take?
• [ ] why isn't coherence a matrix?
• [ ] so, do we understand robust pca? let's incorporate?
• [ ] instructions for formatting token?
• [ ] sparklines can be closer vertically to one another
• [ ] how does random svd work?
• [ ] how much error does power iteration induce?