pathanal.R

pathanal <-function(mcf_gadata,dimension="basicChannelGroupingPath"){
  mcf_gadata <-mcf_gadata %>% 
    mutate(totalConversions=as.numeric(totalConversions))

  path <- with(mcf_gadata,sapply(eval(parse(text=dimension)),function(x)strsplit(x," > ")))
  
  node <-unname(unlist(path))
  node <-gsub("CLICK|:|NA","",node)
  pathlength <- sapply(path,length)
  n <- length(path)
  m <- length(node)
  freq <- a <- d <-numeric(m)
  d[cumsum(pathlength)] <- 1
  a[c(1,which(d==1)+1)[-(n+1)]] <- 1
  
  freq <- with(mcf_gadata,unlist(mapply(rep,totalConversions,each=pathlength)))
  dat <-data_frame(node,a,d,freq)
  
  harvest <- dat %>% 
    group_by(node) %>% 
    summarise(harv_prob=sum(d*freq)/sum(freq))
  
  TotalCV <-with(mcf_gadata,sum(totalConversions))
  
  firsttouch <- dat %>% 
    group_by(node) %>% 
    summarise(first_prob=sum(a*freq)/TotalCV) 
  
  trans_mat <-dat %>% 
    mutate(to=lead(node)) %>% 
    group_by(node,to) %>% 
    summarise(freq=sum((1-d)*freq)) %>% 
    dplyr::filter(!is.na(to) & freq>0) %>% 
    rename(from=node) %>% 
    group_by(from) %>% 
    mutate(trans_prob=freq/sum(freq)) %>% 
    select(-freq) %>% 
    ungroup()
  
  null_node <-group_by(dat) %>% 
    group_by(node) %>% 
    mutate(ind=all(a==1) & all(d==1)) %>% 
    dplyr::filter(ind) %>% 
    select(node,freq) %>% 
    ungroup()
  
  list(unique_node=unique(dat$node),trans_mat=trans_mat,null_node=null_node,
       harvest=harvest,firsttouch=firsttouch)
}

simpath <- function(n,K,trans_prob,harv_prob,first_prob){
  out <-vector("list",n)
  path <- numeric(0)
  j=1
  tmp <- sample.int(K,1,prob = first_prob)
  while(j<=n){
    path <- c(path,tmp)
    if(runif(1)<harv_prob[tmp]){
      out[[j]] <- path
      path <- numeric(0)
      tmp <-  sample.int(K,1,prob = first_prob)
      j=j+1
    }else{
      tmp <-sample.int(K,1,prob = trans_prob[tmp,])
    }
  }
  data_frame(path=sapply(out,paste,collapse=" > ")) %>% 
    group_by(path) %>% 
    summarise(totalConversions=n())
}