R中以日期为变量的Apriori（关联规则学习）？

Question

Is it possible to include a date variable when mining association rules using an algorithm like apriori? Say an "invoice ID" isn't available to clearly define a customer purchase set, but we still have an email address. If Bob bought milk, and 6 months later Bob bought rice, there shouldn't be an association rule (or maybe an association rule with a weak support) for milk and rice. However, if bob buys milk and a day later buys rice, there should be an association rule (and the support should reflect the proximity of dates).

Answer 1

If you don't insist on apriori , have a look at

library(arulesSequences)
lst<- list(
  "bob"=c("milk", "cacao", "beer", "diapers"),
  "lily"=c("flowers", "card")
)
trans <- as(as.list(unlist(lst)), "transactions")
transactionInfo(trans)$sequenceID <- rep(names(lst), lengths(lst))
transactionInfo(trans)$eventID <- as.integer(Sys.Date() + c(0, 6, 60, 60+6*30, 0, 31))
s <- cspade(trans, param=list(maxgap=30))
inspect(s[size(s)>1])
 #   items     support 
 # 1 <{milk},   
 #    {cacao}>     0.5 

maxgap wants 30 days in the example. You may want to adjust that.