In some situations, large objects that are removed to free up memory are needed later in the program. R provides tools to save data to the disk and reload them later when enough memory is available. Returning to the retail sales data example, suppose that we need the sales.data data frame for further processing after mining for frequent itemsets. We can save it to the disk using saveRDS() and reload it later using readRDS():

trans.list <- split(sales.data$item, sales.data$trans)
saveRDS(sales.data, "sales.data.rds")
rm(sales.data)
trans.arules <- as(trans.list, "transactions")
rm(trans.list)
freq.itemsets <- apriori(trans.arules, list(support = 0.3))
sales.data <- readRDS("sales.data.rds")
# Perform further processing with sales.data

The saveRDS() and readRDS() functions save one object at a time without the name of the object. For example, the name sales.data is not saved. However, the column names trans and items are saved. As an alternative...