Open AccessA study on website log data analysis methodology by transition probability
Author(s) -
Jae Kyeong Lee,
Mi Hwan Hyun,
Dong Gu Shin
Publication year - 2018
Publication title -
international journal of engineering and technology
Language(s) - English
Resource type - Journals
ISSN - 2227-524X
DOI - 10.14419/ijet.v7i2.12.11118
Subject(s) - occupancy , stochastic matrix , computer science , markov chain , web analytics , data mining , web page , matrix (chemical analysis) , analytics , transition (genetics) , statistics , data science , world wide web , mathematics , engineering , machine learning , web development , biochemistry , chemistry , gene , architectural engineering , materials science , web application security , composite material
Background/Objectives: To measure occupancy using transition probability matrix as a data analysis method to predict future requirements for web use. From this study, Executives facing business challenges can enhance the decision-making process for management and can be provided quantified evidence.Methods/Statistical analysis: Transition matrix and transition probability matrix are estimated if web users’ webpage use patterns are tied with frequency, using web log data. Occupancy is forecasted based on a Markov chain model.Findings: Data analysis from the perspective of web log-based marketing mostly focuses on increasing traffic and improving transition rates. However, general-purpose tools such as Google Analytics provide diverse web log data. In assumption of independence on users’ page reload, occupancy can be easily estimated through matrix on page reload (transition). As a result, we obtained slightly different results from the usual method that reported only frequency. In particular, rather than making business decisions with the frequency of absolute concepts, we were able to identify the top priority services through the percentage value of relative concepts.Improvements/Applications: The occupancy prediction using transition matrix is about future prediction based on previous information. However, it differs from marketing techniques in that it is estimated based on probability. In addition, it is able to predict more accurately through a probability model.