We currently have the minCharacterizationMean which is really useful for reducing the number of covariates when running an aggregate analysis. I generally set this value of 1% and that returns a smaller subset of useful features. Sometimes I have small cohorts (<100) and then the threshold of 1% doesn't filter anything. It would be useful to have a minCharacterizationCount in addition to minCharacterizationMean where for smaller cohorts a minCharacterizationCount of 2 will remove covariates than only one person has in the smaller cohorts to reduce the set of covariates.
We currently have the minCharacterizationMean which is really useful for reducing the number of covariates when running an aggregate analysis. I generally set this value of 1% and that returns a smaller subset of useful features. Sometimes I have small cohorts (<100) and then the threshold of 1% doesn't filter anything. It would be useful to have a minCharacterizationCount in addition to minCharacterizationMean where for smaller cohorts a minCharacterizationCount of 2 will remove covariates than only one person has in the smaller cohorts to reduce the set of covariates.