• Journal of Korea Society of Industrial Information Systems
• /
• v.20 no.2
• /
• pp.1-9
• /
• 2015

In most techniques of information differentiating for data stream mining, they give larger weight to the information generated in recent compared to the old information. However, there can be important one among the old information. For example, in case of a person was a regular customer in a retail store but has not come to the store in recent, old information with the shopping record of the person can be importantly used in a target marketing for increasing sales. In this paper, highly attention itemsets(HAI) are defined, which mean the itemsets generated in the past frequently but not generated in recent. In addition, a twao-way decay mechanism and a data stream mining method for finding HAI are proposed.

• The Journal of the Korea Contents Association
• /
• v.8 no.11
• /
• pp.347-355
• /
• 2008

This research has been conducted against Cheong-Ju undergraduate students. A total of 213 questionnaires was used for the data. The first purpose of this research is to analyze eating habits of food-related lifestyle of the students. The second purpose is to classify them into groups depending on their food-related lifestyle. The third purpose of this research is to identify eating out behavior patters among the groups classified by food-related lifestyle. Based on the information obtained from those results, this study would help restaurants have their marketing strategies. From the factor analysis, 6 factors of "popular-oriented group", "health-oriented group", "safety-oriented group", "decoration-oriented group", "tasted-oriented group" and "wellbeing-oriented group" were extracted. Additionally, "safety type group", "passive food groups" and "high-interest in foods group" were classified from the cluster analysis.