Hello from ODS. We responded to the idea of tutu.ru to work with their dataset of the passenger flow of the Russian Federation. And if in a post Milfgard huge table of conclusions and scientific, then we want to tell what's under the hood.
What, again, another post about COVID-19? Yes, but no. It was interesting to us from the point of view of mathematical methods and working with an interesting data set. Before you see beautiful pictures and graphs under the cut, I must say a few things:
- any simulation is a very complex process, inside of which there is an incredible amount of IF and ASSUME. We will talk about them.
- those who worked on this article are not epidemiologists or virologists. We are just a group of graph theory enthusiasts who practice methods of modeling complex systems. It is funny, but it is in bioinformatics that the most significant progress of this narrow area of mathematics is now taking place. Therefore, we understand the language of biologists, although we do not know how to justify epidemiological models and make medical conclusions.
- . , . .
- tutu.ru, .
- ODS.ai ML for Social Good ( #ml4sg ODS) , . , , TODO.
— - .
, , .
(safronov).
, :
- 5 , : , -, , , .
- 26 9736 ( ) 80% , — 951 .
: . , — .
TODO: , - , . , .
: SIR. . , - , .
( ):
contact_rate . -> .recovery_rate — , .
" " — " " :
- 0 .
- 1 ( SIR-).
- .
- , 2.
- . - , .
1000 . .
24 2020 .
, , .
- , , contact_rate, .
- . , , .
- . .
- , . .
TODO: . .e. , , / . 2 98 -> 10 90 . , - , .
:
.
TODO: .
: , .
, .
contact_rate = 0.27
recovery_rate = 0.05
: , , . .
: .
( contact_rate), 10% .
contact_rate = 0.21
recovery_rate = 0.05
: , . , . , . , , . , .
, , . .
. .
ODS , -, #ml4sg. .
.
. ODS.