read.R

library(plyr)

synthPop <- read.table("../../newSample100.txt", sep = "\t", header = TRUE)

breaks <- c(0, which(diff(synthPop$hhid) <= 0), nrow(synthPop))
seq_lengths <- diff(breaks)
type_levels <- rep("other", length(seq_lengths))
type_levels[1:3] <- c("owner", "spouse", "child")
synthPop$agent_type <- factor(rep(type_levels, seq_lengths), levels = unique(type_levels), ordered = TRUE)
synthPop <- arrange(synthPop, hhid, agent_type)
xtabs(~sex+agent_type, synthPop)

##      agent_type
## sex    owner spouse  child  other
##   f   291830    744  37927   7752
##   m   771500 371570 695563 783101
##   XXX      0 576708    565    139

xtabs(~age+agent_type, synthPop)

##           agent_type
## age         owner spouse  child  other
##   age04         0    744  37933   7790
##   age09         0  70566  49788   4003
##   age14       109 527778  47964  73669
##   age19      5936 344066  11834  25350
##   age24     30717   1599   2691   2107
##   age29     84890     45    200   3317
##   age34    124625     43   1106  28786
##   age39    143631     85    451  20481
##   age44    167951    730     32 587279
##   age49    156899    941 272803   8439
##   age54    139805    156  58585    954
##   age59     75997    161 198433  24966
##   age64     49785   1005  50334    376
##   age65_up  82985   1103   1901   3475