A Resamping Approach for Customer Gender Prediction Based on E-Commerce Data

Abstract

Demographic attributes of customers such as gender, age, etc. provide the important information for e-commerce service providers in marketing, personalization of web applications. However, the online customers often do not provide this kind of information due to the privacy issues and other reasons. In this paper, we proposed a method for predicting the gender of customers based on their catalog viewing data on e-commerce systems, such as the date and time of access, the products viewed, etc. The main idea is that we extract the features from catalog viewing information and employ the classification methods to predict the gender of the viewers. The experiments were conducted on the datasets provided by the PAKDD’15 Data Mining Competition and obtained the promising results with a simple feature design, especially with the Bayesian Network method along with other supporting techniques such as resampling, cost-sensitive learning, boosting etc.

Building Semantic User Profile for Polish Web News Portal

“…Demography prediction has also been explored in the context of other domains such as e-commerce [9] and authorship profiling (predicting features from usergenerated texts). An overview of the Author Profiling Task at PAN 2016 with the objective of gender and age prediction in the cross-genre perspective was presented in [32].…”

Inferring gender and age of customers in shopping malls via indoor positioning data

“…However, prior knowledge is often required in this method. Feature-based classification methods, such as Bayesian networks (Hu et al., 2007), random forest (RF) (De Bock and Van den Poel, 2010), and other machine learning models (Duc et al., 2017), have been proposed to overcome this weakness by directly inferring a customer profile based on the features of the customer. The features that are considered include browsing behavioral characteristics (i.e.…”

“…the duration of the session, the time of the day, the day of the week, and the number of products viewed) and product-based information (i.e. the view frequency of the products and the tendency of different customers to view different categories) (Duc et al., 2017; Lu et al., 2015; Phuong, 2014).…”