Thesis Details
Mining Multi-Level Sequential Patterns
Mining sequential patterns is a very important area of the data mining. Many industrial and business applications save sequential data where the ordering of transactions is defined. It can be used for example for analysis of consecutive shopping transactions.This thesis deals with the using of concept hierarchies of items for mining sequential patterns. This thesis focuses on two basic approaches - mining level-crossing sequential patterns and mining multi-level sequential patterns. The approaches for the both data mining tasks are formalized and there are proposed data mining algorithms hGSP and MLSP to solve these tasks. Experiments verified that mainly the MLSP has good performance and stability. The usability of newly obtained patterns is shown on the real-world data mining task.
data mining, mining sequential patterns, concept hierarchies, closed patterns, level-crossing sequential patterns, multi-level sequential patterns
@phdthesis{FITPT884,
author = "Michal \v{S}ebek",
type = "Ph.D. thesis",
title = "Mining Multi-Level Sequential Patterns",
school = "Brno University of Technology, Faculty of Information Technology",
year = 2017,
location = "Brno, CZ",
language = "english",
url = "https://www.fit.vut.cz/study/phd-thesis/884/"
}