search

Datawheel / canon-stats

Reusable statistical and complexity calculations using NodeJS and Express
1 stars 1 forks source link

[eci endpoint] fix subnational method calculation #34

Closed AlanORB closed 3 years ago

AlanORB commented 3 years ago

image Lately, we have been noticing that the ECI values for subnational cubes aren't doing great and this might be a signal that the subnational method has a bug

AlanORB commented 3 years ago

for instance, for russia all values go between 0.4 and -1.2, which dont make sense

davelandry commented 3 years ago

😬 @AlanORB any idea who would be the best person to fix this?

AlanORB commented 3 years ago

@davelandry I'm talking with marcos now

mcperez2 commented 3 years ago

@AlanORB can you close this issue with a summary about the problem?

AlanORB commented 3 years ago

For smaller countries the ECI endpoint wasn't giving us results that seemed fine, such as Aysen as the most complex region of Chile, we met with Cesar and Felipe to validate the results and came into conclusion that it was the quality of data which gave us bad results, considering that for bigger countries the results were looking good.