Uma maneira de fazer isso é usar a função 'impute' do pacote R 'Hmisc' e 'rxDataStep'.
Aqui está um exemplo de código R para impute o custo médio por declaração:
car.insurance.df <- data.frame(AgeLevel=factor(c(rep("17-20",4), rep("21-24",4),rep("25-29",4),rep("30-34",4),
rep("35-39",4),rep("40-49",4),rep("50-59",4),
rep("60+",4))), Car.group = factor(rep(c("A","B","C","D"),8)),
Veh.Age=factor(rep(c(rep("0-3",4),rep("4-7",4),rep("8-9",4),
rep("10+",4)),2)),
AvgCost.claim = c(289,372,189,763,282,249,288,850,133,288,179,
NA,160,11,NA,NA,302,420,268,407,194,243,343,320,135,196,293,205,166,135,104,NA),
claims.weights = c(8,10,9,3,8,28,13,2,4,1,1,0,1,1,0,0,18,59,44,24,31,96,39,18,10,13,7,2,4,3,2,0))
rxDataFrameToXdf(data = car.insurance.df, outFile = "car.insurance.xdf")
install.packages("Hmisc")
library(Hmisc)
rxDataStep(inData = "car.insurance.xdf", outFile = "car.insurance-imputed.xdf", transforms = list(AvgCost.claim.New = impute(AvgCost.claim)),
transformPackages = "Hmisc")
rxSummary(~Car.group + AvgCost.claim, data = "car.insurance-imputed.xdf", fweights ="claims.weights")
