Based Big Data Analysis of Fraud Detection for Online Transaction Orders

Abstract

Abstract. Fraud control is important for the online marketplace. This study addresses the problem of detecting attempts to deceive orders in Internet transactions. Our goal is to generate an algorithm model to detect and prevent the fraudulent orders. First, after analyzing the real historical data of customers' orders from Dangdang Website (http://www.dangdang.com. E-commerce China Dangdang Inc (Dangdang) is a leading e-commerce company in China. Dangdang officially listed on the New York Stock Exchange on December 8th, 2010, and is the first Chinese B2C e-commerce company which is completely based on online business to list on New York Stock.), we described characteristics related to transactions that may indicate frauds orders. We presented fraudulent orders characteristic matrix through comparing the normal and abnormal orders. Secondly, we apply Logic Regression model to identify frauds based on the characteristic matrix. We used real data from Dang company to train and evaluate our methods. Finally we evaluated the validity of solutions though analyzing feedback data.

Fatores de sucesso do big data no negócio eletrónico: revisão narrativa de literatura

“…O Big Data é uma grande oportunidade para os negócios eletrónicos rastrearem o comportamento dos consumidores e determinarem a maneira mais eficaz de transformá-los em clientes fidelizados (Ilieva et al, 2015). Outro aspeto importante do Big Data é a possibilidade de ser feita a personalização de preços, fazendo com que a determinação de um preço "correto" leve em consideração dados de várias fontes como os preços de concorrentes, as vendas de produtos, preferências regionais e o perfil do consumidor (Yang et al, 2014).…”

Leveraging Product Characteristics for Online Collusive Detection in Big Data Transactions

“…Maranzato et al [50] used logistic regression to detect and identify features of credit fraud [57]. Yang et al [73] applied logistic regression model to detect fraud online transaction in Dangdang platform. For fraud detection research, a review by Rana and Baria [37] summarized researches on fraud detection techniques over the past decade.…”

Machine Learning Based Prediction and Prevention of Malicious Inventory Occupied Orders

“…They tried to enhance the characteristics set and adopted a Logistic Regression Model that can deal with this set. Yang et al (2014) extracted the characteristics of cheating orders from the historical data of orders, and uncovered fraudulent orders by applying Logistic Regression algorithm. Some new techniques from other closely related fields, such as system design and performance analysis, can also be adopted to further enhance the research in MIOOs.…”