library(gamlr)

## Browsing History. 
## The table has three colums: [machine] id, site [id], [# of] visits
web <- read.csv("CS2006domains.csv")

## Tell R that 'id' is a factor; we know there are 10000 machines
n <- 1e4
web$id <- factor(web$id, levels=1:n)

## Read in the actual website names, and use these to create a site factor
## We know that there are 1000, and these are their names (in correct order)
d <- 1e3
sitenames <- scan("CS2006sites.txt", what="character")
web$site <- factor(web$site, levels=1:d, labels=sitenames)

## get total visits per-machine and % of time on each site
## tapply(a,b,c) does c(a) for every level of factor b.
machinetotals <- as.vector(tapply(web$visits,web$id,sum)) ## it returns matrix; we'll make it a vector
visitpercent <- 100*web$visits/machinetotals[web$id]

## use this info in a sparse matrix
## this is something you'll be doing a lot; familiarize yourself.
xweb <- sparseMatrix(
	i=as.numeric(web$id), j=as.numeric(web$site), x=visitpercent,
	dims=c(nlevels(web$id),nlevels(web$site)),
	dimnames=list(id=levels(web$id), site=levels(web$sites)))

## transactions
spend <- read.csv("CS2006transactions.csv")
spend$product <- factor(spend$product, levels=1:d)
spend$id <- factor(spend$id, levels=1:n)
xspend <- sparseMatrix(
	i=as.numeric(spend$id), j=as.numeric(spend$product), x=spend$dollars, 
	dims=c(nlevels(spend$id),nlevels(spend$product)),
 	dimnames=list(id=levels(spend$id), product=levels(spend$product)))

all(rownames(xspend)==rownames(xweb))

yspend <- cbind(id=as.numeric(rownames(xspend)),spend=rowSums(xspend))
write.table(yspend, row.names=FALSE, file="CS2006totalspend.csv", sep=",",quote=FALSE)

## demographics
demo <- read.csv("CS2006demographics.csv")
demo$id <- factor(demo$id)
Xdemo <- sparse.model.matrix(id ~ .-1, data=demo)