colnames(i_r_v_c_cat) <-c("base", "reco", "view", "click", "base_cat", "reco_cat")
Filter columns of interest
i_r_v_c_cat <- i_r_v_c_cat[i_r_v_c_cat$base_cat %in% fil,]
i_r_v_c_cat <- i_r_v_c_cat[i_r_v_c_cat$reco_cat %in% fil,]
#Add a ctr columns
i_r_v_c_cat$ctr <- round(i_r_v_c_cat$click*100/i_r_v_c_cat$view, 2)
#filter out items whose ctr score is greater than 1
i_r_v_c_cat<- i_r_v_c_cat[i_r_v_c_cat$ctr<100.0, ]
#Compute the base category, recommendation category ctr heat map
cat_base_reco <-i_r_v_c_cat[, c("base_cat","reco_cat","view","click")]
cat_base_reco<-aggregate(. ~base_cat+reco_cat, data=cat_base_reco, FUN=sum)
cat_base_reco$ctr <- round(cat_base_reco$click*100/cat_base_reco$view,2)
resultsMatrix <- matrix(0, length(fil), length(fil))
fil ->colnames(resultsMatrix) ->rownames(resultsMatrix)
for (i in 1:100){
resultsMatrix[as.vector(cat_base_reco[["base_cat"]][i]), as.vector(cat_base_reco[["reco_cat"]][i])] <-as.vector(cat_base_reco[["ctr"]][i])
}
write.table(resultsMatrix, file="../data/sigir2016short/output/category_heat_map.txt", sep="&")
#Compute Base category ctr
cat_base<-aggregate(. ~base_cat, data=cat_base_reco, FUN=sum)
cat_base$ctr <- round(cat_base$click*100/cat_base$view,2)
write.table(cat_base, file="../data/sigir2016short/output/category_ctr.txt", sep="&")
#Compute base item recommendation item ctr
i=250
top_base_reco <- merge(i_r_v_c_cat[order(-i_r_v_c_cat$view),][1:i,], i_r_v_c_cat[order(-i_r_v_c_cat$click),][1:i,], by =c("base","reco"), all=F)
top_base_reco <- top_base_reco[, c("base","reco","view.x","click.x","base_cat.x","reco_cat.x","ctr.x")]
write.table(top_base_reco, file="../data/sigir2016short/output/top_base_reco.txt", sep="&")
i=4800
bot_base_reco <- merge(i_r_v_c_cat[order(-i_r_v_c_cat$view),][1:i,], i_r_v_c_cat[order(i_r_v_c_cat$click),][1:i,], by =c("base","reco"), all=F)
bot_base_reco <- bot_base_reco[, c("base","reco","view.x","click.x","base_cat.x","reco_cat.x","ctr.x")]
write.table(bot_base_reco, file="../data/sigir2016short/output/bot_base_reco.txt", sep="&")
#Compute Base item ctr
i=72
i_v_c_cat<-i_r_v_c_cat[, c("base", "base_cat","view","click"),]
i_v_c_cat <-aggregate(. ~base+base_cat, data=i_v_c_cat, FUN=sum)
i_v_c_cat$ctr<-round(i_v_c_cat$click*100/i_v_c_cat$view, 2)
top_base <- merge(i_v_c_cat[order(-i_v_c_cat$view),][1:i,], i_v_c_cat[order(-i_v_c_cat$click),][1:i,], by =c("base","base_cat"), all=F)
top_base <-top_base[, c("base","base_cat","view.x","click.x", "ctr.x")]
write.table(top_base, file="../data/sigir2016short/output/top_base.txt", sep="&")
#Bottom scoring political items
i=600
i_v_c_cat <- i_v_c_cat[i_v_c_cat$base_cat=="politik", ]
bot_base <- merge(i_v_c_cat[order(-i_v_c_cat$view),][1:i,], i_v_c_cat[order(i_v_c_cat$click),][1:i,], by =c("base","base_cat"), all=F)
bot_base <-bot_base[, c("base","base_cat","view.x","click.x", "ctr.x")]
write.table(bot_base, file="../data/sigir2016short/output/bot_base.txt", sep="&")