E-commerce transactions of behaviors anomaly detection system

Abstract EN

This article proposes designing a system to detect anomalies for e-commerce sites.

The proposed system consists of two phases the first clustering phase and the second classification phase, in the first stage, the algorithm was proposed modify k-means (PMK-means) algorithm as clustering phase and support vector machine (SVM) algorithm as classification phase to solve the problems and abnormal transactions of e-commerce data.

This proposed modify kmeans will be applied as a preprocessing stage, the main objective is to generate a collection of clusters from the ecommerce transactions of anomaly dataset.

Then apply the SVM algorithm to classifications the dataset and anomaly detection.