Apologies for a very naive question. I have a population of some 10,000 observations and 10 variables. From the 10 variables I have derived a single (and meaningful) continuous metric. With missing values I am using multiple imputation to analyse the entire data set.
In the set I have two well defined classes, say class A, with 7000 members, and class B with 2000 members. And I have the derived metric for each observation.
I then have a population C of say 200 observations derived from the 1000 uncategorized observations for which the categorization into class A or B in unknown. I wish to estimate the probability that the population C is actually drawn from population A rather than population B (and v.v) based on the derived metric.
What is an appropriate approach?
In the set I have two well defined classes, say class A, with 7000 members, and class B with 2000 members. And I have the derived metric for each observation.
I then have a population C of say 200 observations derived from the 1000 uncategorized observations for which the categorization into class A or B in unknown. I wish to estimate the probability that the population C is actually drawn from population A rather than population B (and v.v) based on the derived metric.
What is an appropriate approach?
