AJ estimate from performance data is more complicated than the binary case.
In _create_rtichoke_curve_list_binary() I use aj_estimates_from_performance_data for:
- Extract use case: a) One Model b) Several Models c) Several Populations
- Create reference lines.
In order to find a good solution I need to figure out what aj estimate is being used when I create reference lines, and if possible to extract_aj_estimate for a specific cutoff.
AJ estimate from performance data is more complicated than the binary case.
In
_create_rtichoke_curve_list_binary()I useaj_estimates_from_performance_datafor:In order to find a good solution I need to figure out what aj estimate is being used when I create reference lines, and if possible to extract_aj_estimate for a specific cutoff.