|
Title:
|
DISTRIBUTED TREE-BASED DECISION RULE LEARNING |
|
Author(s):
|
?ukasz Wróbel and Wojciech Sikora |
|
ISBN:
|
978-989-8533-95-1 |
|
Editors:
|
Hans Weghorn |
|
Year:
|
2019 |
|
Edition:
|
Single |
|
Keywords:
|
Rule Learning, Decision Rules, Classification, Big Data, Apache Spark |
|
Type:
|
Full Paper |
|
First Page:
|
47 |
|
Last Page:
|
53 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
The paper describes a decision rule induction algorithm for a distributed data environment. Recent advances in
processing a large volume of data result in the rapid development of many machine learning algorithms adapted to work
in such systems. However, there is still a relatively small number of distributed decision rule learning algorithms. Much
more attention has been paid to other types of machine learning techniques, such decision trees. In comparison to trees, a
rule-based model can be however simpler and thus more comprehensible for humans while preserving similar predictive
accuracy as trees.
The paper evaluates a simple approach to rule learning in a distributed environment, utilizing a decision tree as an inner
learner in separate-and-conquer fashion. The effectiveness of the algorithm was empirically tested on real and artificial
benchmark datasets in the Spark environment. The results show that the proposed algorithm can induce compact models
of superior accuracy and comprehensibility. |
|
|
|
|
|
|
