如何在R中对顺序分类数据进行聚类

Question

Consider a data set where users can choose among 3 activities, and we have the data for the choice of their first 10 activities. Example data:

for (i in 1:10) 
{
  # sample from list of 3 strings using a set probability
  x <- sample( c("A", "B", "C"), 1000, replace=TRUE, prob=c(0.5, 0.3, 0.2) )
  # assign to variable created on the fly
  assign( paste("cat", i, sep=""), x )
}

first10 <- data.frame(cat1, cat2, cat3, cat4, cat5, cat6, cat7, cat8, cat9, cat10)

What's the best approach in R to cluster users according to their activity sequence?

I've looked around on stackoverflow, and the most similar questions ask about how to cluster categorical data in R (which is part of the analysis), but this in and of itself doesn't account for the sequential nature of the data. Are there R packages that are well-suited for this analysis?

Answer 1

Look for frequent itemset mining instead of clustering.

Most clustering methods are for continuous numerical data, and assume some vector field. They take every position into account.

A frequent pattern, however, may be only part if a sequence, a sequence may exhibit multiple (or none) of these patterns, and patterns may have gaps inbetween. All of these properties are usually desirable.