Meeting Notes March 4/14, 2019:

Pretty much skipped these first 3 sections since we finished almost all of our V01 tickets. See action Items for planning.

Demo:

what should we make for V0.2?
- classify heart rate activity to particular medical conditions (need annotated data)
- one model to predict rates, and another to predict classes based on those predicted rates
- color classifications in series
- regression + classification (predict day vs partial day)
- classify some sleep parameter from our existing data (e.g., sleep score)
- color prediction line with classification color
- real-time monitoring (intermittent online learning, continuous predictions)
- incorporate other inputs
- enable more users
- enable more users with their own data
- multiple days listed in multiple charts
- next/prev day
- database
- 3 charts - one per model, plus 1 showing all models together
what are criteria for a good choice (of what to make)?
- usefulness in jobs
- usefulness in interviews
- fun
- interesting
- working code
- accomplishable in 2 weeks
what will we make?
- heroku #161
- integrating database #167
- users (oauth, routes) #180
- classification (selecting broader piece of sleep data to get e.g. score) #179

[x] create tickets for the things we'll make
[x] api readme
[x] MS3 instructions for thomas to run tests similar to how travis does it
[x] MS3 determine if thomas needs to run ui, and if so, he should be able to with the test install, plus pawgo (the expanded version)
[ ] ask thomas if we need to make the things we have user test cases for
[x] Let ben know following user test cases
[x] user test cases
1. user can log in with oauth (like we do with our google accounts on moodle)
2. word "Loading..." appears until chart renders
3. successful initial chart render
brian walk us through ML code next week (RNN, then ODE)