Closed robgur closed 1 month ago
OK, i like the setup where we call the field "certainty" with available values as "low" and "high" -- that is very clear what we are talking about.
Agreed on "... will also have to change the field 'proportion_equivocal_family' to 'proportion_low_certainty_family', and 'accuracy_excluding_equivocal_family' to 'accuracy_excluding_low_certainty_family'"
Thoughts @robgur ?
OK, i like the setup where we call the field "certainty" with available values as "low" and "high" -- that is very clear what we are talking about.
Agreed on "... will also have to change the field 'proportion_equivocal_family' to 'proportion_low_certainty_family', and 'accuracy_excluding_equivocal_family' to 'accuracy_excluding_low_certainty_family'"
Thoughts @robgur ?
I am in agreement on all this.
Closing issue. columns.csv and certainty.csv files updated with text suggested here in latest commit.
From Russell: A bit late to this, but I'm okay with equivocal field being called 'certainty', but I'm not sure I am comfortable calling a model prediction 'certain'. What if the values for certainty were just 'high' and 'low'? Equivocal is low certainty, Unequivocal is high certainty.
I suppose we will also have to change the field 'proportion_equivocal_family' to 'proportion_low_certainty_family', and 'accuracy_excluding_equivocal_family' to 'accuracy_excluding_low_certainty_family'?