Caros, rotineiramente me deparo com um problema que não sei como resolver. Considerem o seguinte exemplo: (c1 <- trunc(runif(1000,1,10))) (c2 <- trunc(runif(1000,5,70))) (c3 <- trunc(runif(1000,1,4))) (var <- rnorm(1000,10,2)) (peso <- round(rnorm(1000,1,.2)^2,1)) df <- data.frame(c1,c2,c3,var,peso) Gostaria de uma função que pudesse me fornecer em uma única tabela do tipo cols = c1,c2,c3,count,medvar e, mais genericamente, cols = c1,c2,c3,somapeso,medpondvar com todas as combinações existentes (as vezes todas as teoricamente possiveis) entre c1,c2,c3 c1 c2 c3 c1*c2 c1*c3 c2*c3 c1*c2*c3 De forma mais específica, gostaria: t1 <- aggregate(list(count = df$var), list(c1=rep('*tot',1000) ,c2=rep('*tot',1000),c3=rep('*tot',1000)),length) tc1 <- aggregate(list(count = df$var), list(c1=df$c1 ,c2=rep('*tot',1000),c3=rep('*tot',1000)),length) tc2 <- aggregate(list(count = df$var), list(c1=rep('*tot',1000),c2=df$c2 ,c3=rep('*tot',1000)),length) tc3 <- aggregate(list(count = df$var), list(c1=rep('*tot',1000),c2=rep('*tot',1000),c3=df$c3 ),length) tc1c2 <- aggregate(list(count = df$var), list(c1=df$c1 ,c2=df$c2 ,c3=rep('*tot',1000)),length) tc1c3 <- aggregate(list(count = df$var), list(c1=df$c1 ,c2=rep('*tot',1000),c3=df$c3 ),length) tc2c3 <- aggregate(list(count = df$var), list(c1=rep('*tot',1000),c2=df$c2 ,c3=df$c3 ),length) tc1c2c3 <- aggregate(list(count = df$var), list(c1=df$c1 ,c2=df$c2 ,c3=df$c3),length) comp1 <- rbind(t1,tc1,tc2,tc3,tc1c2,tc1c3,tc2c3,tc1c2c3) rm(t1,tc1,tc2,tc3,tc1c2,tc1c3,tc2c3,tc1c2c3) t1 <- aggregate(list(medvar = df$var), list(c1=rep('*tot',1000) ,c2=rep('*tot',1000),c3=rep('*tot',1000)),mean) tc1 <- aggregate(list(medvar = df$var), list(c1=df$c1 ,c2=rep('*tot',1000),c3=rep('*tot',1000)),mean) tc2 <- aggregate(list(medvar = df$var), list(c1=rep('*tot',1000),c2=df$c2 ,c3=rep('*tot',1000)),mean) tc3 <- aggregate(list(medvar = df$var), list(c1=rep('*tot',1000),c2=rep('*tot',1000),c3=df$c3 ),mean) tc1c2 <- aggregate(list(medvar = df$var), list(c1=df$c1 ,c2=df$c2 ,c3=rep('*tot',1000)),mean) tc1c3 <- aggregate(list(medvar = df$var), list(c1=df$c1 ,c2=rep('*tot',1000),c3=df$c3 ),mean) tc2c3 <- aggregate(list(medvar = df$var), list(c1=rep('*tot',1000),c2=df$c2 ,c3=df$c3 ),mean) tc1c2c3 <- aggregate(list(medvar = df$var), list(c1=df$c1 ,c2=df$c2 ,c3=df$c3),mean) comp2 <- rbind(t1,tc1,tc2,tc3,tc1c2,tc1c3,tc2c3,tc1c2c3) rm(tc1,tc2,tc3,tc1c2,tc1c3,tc2c3,tc1c2c3) comp <- merge(comp1,comp2,by=c('c1','c2','c3'),all=T) rm(comp1,comp2) Exite algum pacote que faca isso?