You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I noticed that the most likely model is highlighted quite a lot, even if another model is very close to it.
I find especially the tick mark ✅ a bit misleading: here there's no overwhelming evidence it was a chronic infection. It's just compatible with it.
This might cause people to make false positive deductions.
for example, chronic infection might be much less likely to occur in general (lower prior) and here the advantage for chronic is very small, yet the result is "it's chronic"
It might make sense to present the output as a pseudo probability table. Rather than saying: this is the winner, say: chronic: 80%, omicron: 19%, molnu: 1%, pre-omicron: 0%. This presentation would provide more context as to the certainty (even if exact probablilities of course can't be trusted as they heavily depend on priors and model assumptions)
The text was updated successfully, but these errors were encountered:
I noticed that the most likely model is highlighted quite a lot, even if another model is very close to it.
I find especially the tick mark ✅ a bit misleading: here there's no overwhelming evidence it was a chronic infection. It's just compatible with it.
This might cause people to make false positive deductions.
for example, chronic infection might be much less likely to occur in general (lower prior) and here the advantage for chronic is very small, yet the result is "it's chronic"
It might make sense to present the output as a pseudo probability table. Rather than saying: this is the winner, say: chronic: 80%, omicron: 19%, molnu: 1%, pre-omicron: 0%. This presentation would provide more context as to the certainty (even if exact probablilities of course can't be trusted as they heavily depend on priors and model assumptions)
The text was updated successfully, but these errors were encountered: