A comparative study of linear and nonlinear models for aggregate retail sales forecasting

摘要: Abstract The purpose of this paper is to compare the accuracy various linear and nonlinear models for forecasting aggregate retail sales. Because strong seasonal fluctuations observed in sales, several traditional methods such as time series approach regression with dummy variables trigonometric functions are employed. versions these implemented via neural networks that generalized functional approximators. Issues modeling deseasonalization also investigated. Using multiple cross-validation samples, we find able outperform their counterparts out-of-sample forecasting, prior adjustment data can significantly improve performance network model. overall best model built on deseasonalized data. While be useful developing effective predicting may not robust. Furthermore, sales forecasting.