arules (version 1.7-9)

APappearance-class: Class APappearance --- Specifying the appearance Argument of Apriori to Implement Rule Templates

Description

Specifies the restrictions on the associations mined by apriori(). These restrictions can implement certain aspects of rule templates described by Klemettinen (1994).

Slots

labels

character vectors giving the labels of the items which can appear in the specified place (rhs, lhs or both for rules and items for itemsets). none specifies, that the items mentioned there cannot appear anywhere in the rule/itemset. Note that items cannot be specified in more than one place (i.e., you cannot specify an item in lhs and rhs, but have to specify it as both).

default

one of "both", "lhs", "rhs", "none". Specified the default appearance for all items not explicitly mentioned in the other elements of the list. Leave unspecified and the code will guess the correct setting.

set

used internally.

items

used internally.

Objects from the Class

If appearance restrictions are used, an appearance object will be created automatically within the apriori() function using the information in the named list of the function's appearance argument. In this case, the item labels used in the list will be automatically matched against the items in the used transactions.

Objects can also be created by calls of the form new("APappearance", ...). In this case, item IDs (column numbers of the transactions incidence matrix) have to be used instead of labels.

Coercions

• as("NULL", "APappearance")

• as("list", "APappearance")

Author

Michael Hahsler and Bettina Gruen

Details

Note that appearance is only supported by the implementation of apriori().

References

Christian Borgelt (2004) Apriori --- Finding Association Rules/Hyperedges with the Apriori Algorithm. https://borgelt.net/apriori.html

M. Klemettinen, H. Mannila, P. Ronkainen, H. Toivonen and A. I. Verkamo (1994). Finding Interesting Rules from Large Sets of Discovered Association Rules. In Proceedings of the Third International Conference on Information and Knowledge Management, 401--407.

See Also

Other mining algorithms: AScontrol-classes, ASparameter-classes, apriori(), eclat(), fim4r(), ruleInduction(), weclat()

Examples

data("Adult")

## find only frequent itemsets which do not contain small or large income
is <- apriori(Adult,
  parameter = list(support = 0.1, target = "frequent"),
  appearance = list(none = c("income=small", "income=large"))
)
itemFrequency(items(is))["income=small"]
itemFrequency(items(is))["income=large"]

## find itemsets that only contain small or large income, or young age
is <- apriori(Adult,
  parameter = list(support = 0.1, target = "frequent"),
  appearance = list(items = c("income=small", "income=large", "age=Young"))
)
inspect(head(is))

## find only rules with income-related variables in the right-hand-side.
incomeItems <- grep("^income=", itemLabels(Adult), value = TRUE)
incomeItems
rules <- apriori(Adult,
  parameter = list(support = 0.2, confidence = 0.5),
  appearance = list(rhs = incomeItems)
)
inspect(head(rules))

## Note: For more complicated restrictions you have to mine all rules/itemsets and
## then filter the results afterwards.