New Rule Induction Method by Use of a Co-occurrence Set from the Decision Table
| Autor: | Tetsuro Saeki, Yuichi Kato |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
050101 languages & linguistics
Relation (database) Association rule learning Computer science business.industry Rule induction 05 social sciences 02 engineering and technology Set (abstract data type) 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing 0501 psychology and cognitive sciences Rough set Artificial intelligence Decision table business Random variable Statistical hypothesis testing |
| Zdroj: | Rules and Reasoning ISBN: 9783030579760 RuleML+RR |
| DOI: | 10.1007/978-3-030-57977-7_4 |
| Popis: | STRIM (Statistical Test Rule Induction Method) has been proposed as an if-then rule induction method from the decision table (DT) and has improved those methods by the conventional Rough Sets from a statistical view. The method recognizes condition attributes (CA) and the decision attribute (DA) in DT as random variables having the causality of an input-output relation, and uses the relation of transforming the inputs (outcomes of CA) into the outputs (those DA) through the rules for rule induction strategies. This paper reconsiders the conventional STRIM, proposes a new rule induction method and strategy named apriori-STRIM and confirms the validity and capacity by a simulation experiment. Specifically, the new method explores CA of causes after receiving outcomes of DA by use of co-occurrence sets of outcomes of CA. The co-occurrence set is a well-known concept in the association rule learning (ARL) field. This paper also clarifies the differences of rule induction methods and their capacities between apriori-STRIM and ARL by the same experiments. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|