Detailed Response Variable Information
This function is called by certain na.action functions if options(na.detail.response=TRUE) is set. By default, this function returns a matrix of counts of non-NAs and the mean of the response variable computed separately by whether or not each predictor is NA. The default action uses the last column of a Surv object, in effect computing the proportion of events. Other summary functions may be specified by using options(na.fun.response="name of function").
na.detail.response(mf)
mf: a model framea matrix, with rows representing the different statistics that are computed for the response, and columns representing the different subsets for each predictor (NA and non-NA value subsets).
Frank Harrell
Department of Biostatistics
Vanderbilt University
na.omit, na.delete, model.frame.default, naresid, naprint, describe
# sex # [1] m f f m f f m m m m m m m m f f f m f m # age # [1] NA 41 23 30 44 22 NA 32 37 34 38 36 36 50 40 43 34 22 42 30 # y # [1] 0 1 0 0 1 0 1 0 0 1 1 1 0 0 1 1 0 1 0 0 # options(na.detail.response=TRUE, na.action="na.delete", digits=3) # lrm(y ~ age*sex) # # Logistic Regression Model # # lrm(formula = y ~ age * sex) # # # Frequencies of Responses # 0 1 # 10 8 # # Frequencies of Missing Values Due to Each Variable # y age sex # 0 2 0 # # # Statistics on Response by Missing/Non-Missing Status of Predictors # # age=NA age!=NA sex!=NA Any NA No NA # N 2.0 18.000 20.00 2.0 18.000 # Mean 0.5 0.444 0.45 0.5 0.444 # # \dots\dots # options(na.action="na.keep") # describe(y ~ age*sex) # Statistics on Response by Missing/Non-Missing Status of Predictors # # age=NA age!=NA sex!=NA Any NA No NA # N 2.0 18.000 20.00 2.0 18.000 # Mean 0.5 0.444 0.45 0.5 0.444 # # \dots # options(na.fun.response="table") #built-in function table() # describe(y ~ age*sex) # # Statistics on Response by Missing/Non-Missing Status of Predictors # # age=NA age!=NA sex!=NA Any NA No NA # 0 1 10 11 1 10 # 1 1 8 9 1 8 # # \dots