A Study on Customized Brand Recommendation based on Customer Behavior for Off-line Shopping Malls*



Kim, Nam-ki

Department of Business Administration
Graduate School, Kyungil University

Supervised by Professor Jeong, Seok-Bong

(Abstract)

With the rapid growth of the Internet over the recent years, there has been a demand for online shopping malls. There are several contributing factors and one of them being able to apply business analytics to online shopping mall. Business analytics allow collection and analysis of data to model consumer shopping behavior, such as preferences, price range, product style, date of purchase, the quantity of purchase. Online shopping mall, therefore, have more adequate conditions for analytics businesses than offline one, and brand-recommend system is a key part of the conditions.
Recently, Amazon, Netflix, Facebook and domestic shopping malls tend to adopt the recommendation system, whereas the offline shopping malls have relatively inferior systems than those malls as for rapid interactions with customers.
This study is about a recommendation system which can be applied to marketing activities in offline shopping mall. It proposes a real-time brand recommendation system using collaborative filtering, which is widely applied for various companies.
The recommendation system is a business model which recommends products that are closely related to customers'' preference. Considering environments of the offline shopping malls, this study is focused on recommendation system about brands which purchased and visited by customers, by using collaborative filtering.
The process of the proposed model, which applies collaborative filtering to offline shopping mall can be separated into two categories; training and apply process. The training process is designed to make the base brand network(BBN) by using historical transaction data. In the apply process, real-time brand recommendation system is set up and is based on behaviors of customers who have bought products on that day.
By finding Precision, Recall and F1 measure, this study evaluates the efficiency of the recommendation system. Brand transaction history from mid-sized department store in South Korea will be used for comparative evaluation.
This study, related to brand recommendation system for offline shopping mall, points out as in the following.
First, brand recommendation system based on customers'' shopping behavior can be effectively applied to offline shopping malls, as so in online one.
Second, a guideline about various parameters for customized brand recommendation system is suggested.