Furniture market demand forecasting using machine learning approaches

Abstract

Researching consumer demand in the market for goods is essential for any business. The purpose of this study is to forecast the demand for furniture from a large manufacturer in Bulgaria. Significant factors influencing customer flow, both in the company and online stores, are investigated. Daily observations for nearly two years are modelled using CART Ensemble with arcing. The constructed models describe the demand for furniture with high goodness-of-fit statistics: coefficient of determination up to 93% and determine the order of the factors influencing the demand.