for october 26

[rmarren1]

@rmarren1

• modify server to get data from a S3 bucket rather than from user upload.
• get group members code working in the pipeline (ICA, Kernel algorithm, PCA)
• modify processing steps for demo (so it doesn't take too long, e.g. I don't think we need to generate all 111 spectrograms, or dump all of the pre-processed data back to the user)
• add other data visualizations to the report generated (coherence if we get to it, this, this)

^ DoD for all is a demo going from data to analysis

@nkumarcc

• code.md for kernel density- Python Notebook Here
  • code.md for bad electrode detection - Python Notebook Here
• integrate kernel density into pipeline
• coherence code.md (reach) Coherence Python Notebook

@mnatenz1

• code.md for ICA

@tsunwong123

• code.md for PCA- Jupyter Notebook Here

[gkiar]

👌

[jovo]

👌

On Thu, Oct 20, 2016 at 5:33 PM, Ryan Marren notifications@github.com
wrote:

@rmarren1 https://github.com/rmarren1
[ ] - modify server to get data from a S3 bucket rather than from user
upload.
[ ] - get group members code working in the pipeline (ICA, Kernel
algorithm, PCA)
[ ] - modify processing steps for demo (so it doesn't take too long, e.g.
I don't think we need to generate all 111 spectrograms, or dump all of the
pre-processed data back to the user)
[ ] - add other data visualizations to the report generated (coherence if
we get to it, this
https://github.com/NeuroDataDesign/orange-panda/blob/master/eeg_data/EEG_Plots/HeatmapFixed2.jpeg,
this
https://github.com/NeuroDataDesign/orange-panda/blob/master/eeg_data/EEG_Plots/freq_plots/A00051826_rest_Cz.png
)

@nkumarcc https://github.com/nkumarcc
[ ] - code.md for kernel density
[ ] - coherence code.md (reach)

@mnatenz1 https://github.com/mnatenz1
[ ] code.md for ICA

@tsunwong123 https://github.com/tsunwong123

• code.md for PCA

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#17, or mute the
thread
https://github.com/notifications/unsubscribe-auth/AACjcsiqdyMLFV1T0Oz00udLrXchrRwsks5q194agaJpZM4KcoQD
.

the glass is all full: half water, half air.
neurodata.io, jovo calendar
https://calendar.google.com/calendar/embed?src=joshuav%40gmail.com&ctz=America/New_York

[jovo]

KDE

• failure cases for kde: mixture of point masses!
• what metric for evaluating kde?
• likelihood isn't bounded above!
• what's "crazy white noise"?
• play with threshold maybe? determine that simulations actually correctly discard "bad" electrodes

Coherence

• coherence should be a matrix comparing pairs of spectral bands?

PCA

• need simulated data where we know the truth.
• metric is non-sense

Web-service

• why not deployed?
• links to images

check out https://github.com/neurodata/checklists/blob/master/time-series.md
(based on https://github.com/neurodata/checklists/blob/master/matrix.md)
see https://github.com/neurodata/meda for implementation of matrix.md