Big Data Analysis for Revenue and Sales Prediction using Support Vector Regression with Auto-regressive Integrated Moving Average

Abstract

In e-commerce industry, customers’ demands get fluctuated throughout the year depending on the purchasing behavior and season. It may be a repetition period in the year, where sales may generally be down, moderate, and whilst some periods are extremely high. Studies reveal that machine learning techniques boosted much e-commerce industry, from supply chain management to business planning. In this paper, a hybrid big data analytical model which integrates Support Vector Regression (SVR) with Auto-Regressive Integrated Mov- ing Average (ARIMA) is proposed to predict product sales and revenues. The simulation results show that the proposed model presents lower relative error rate and higher accuracy that can be utilized for business planning and strategies.