[Work Log] FIRE - cleanup
May 16, 2014
Skype w/ Kobus
Reflecting on immunity histograms I created late last night. In discussion with Kobus, we've decided a few things could help: (a) log-transforming, (b) introducing a scaling factor, and (c) using robust likelihood (e.g. truncated squared error).
Will Skype with Warren on Monday afternoon to hand off the self-report clustering analysis. Will spend this afternoon cleaning up the data and code to make it easier for him to pick it up.
Kobus will likely need my input in the coming weeks to incorporate branching Gaussian processes into an upcoming grant proposal.
Self-report clustering cleanup
Add documentation and READMES for clustering, experiment directory, output files.
Add matlab script to visualize clustering results.
Created reference page for clustering results.
Posted by Kyle Simek