search

mila-iqia / COVI-AgentSim

Covid-19 spread simulator with human mobility and intervention modeling.
Other
20 stars 8 forks source link

Self testing #64

Closed teganmaharaj closed 4 years ago

teganmaharaj commented 4 years ago

Description

Self-testing reorganized into two categories (touches human.py and core.yaml) : P_TEST_SEVERE_OR_SUSPICIOUS: 0.75 P_TEST_OTHER_REASON: 0.05

"Other reason" is uniform random in the population, and captures what was formerly moderate or mild symptoms, additionally captures people getting tested for other reasons (e,g. travelling soon, they're a hypochondriac, etc.).

With only 2 params, it will be easier to tune to match observed testing rates in Montreal.

Type of change

Please delete options that are not relevant.

teganmaharaj commented 4 years ago

image Flow chart

teganmaharaj commented 4 years ago

Link to lucidchart for that flow chart: https://app.lucidchart.com/invitations/accept/cb5e726e-02fc-4b30-9190-5245b232cc09

mweiss17 commented 4 years ago

great -- will review in ~1 hour

mweiss17 commented 4 years ago

Just a small note, to respond to your comment about "tune to what is happening in montreal" I think that the primary problem wrt tuning testing is the scale. We need larger populations in order to have an accurate (smaller) scale of testing.

mweiss17 commented 4 years ago

I think we should hold off on this until prateek's changes are merged and plots generated. This should take 1-2 days. Is this ok? I just don't want to modify the symptoms / testing infra while that PR is stabilizing.

mweiss17 commented 4 years ago

also I found a variable undefined issue that needs to get fixed for the CI to pass. Thanks!

pg2455 commented 4 years ago

What do you think of scoring in test_facility.py? Should that be changed as well?

mweiss17 commented 4 years ago

I believe this was added in a pr by prateek