Intelligent sales forecasting for fashion retailing using harmony search algorithms and extreme learning machines

A hybrid intelligent (HI) model, comprising a data pre-processing component and a HI forecaster, is developed to tackle the medium-term fashion sales forecasting problem. The HI forecaster adopts a novel learning algorithm- based neural network to generate initial sales forecasts and then uses a heuristic fine-tuning process to obtain more accurate forecasts. The learning algorithm integrates an improved harmony search algorithm and an extreme learning machine. Experiments based on real fashion retail data and public benchmark data sets were conducted to evaluate the performance of the proposed model. Results demonstrate that the performance is far superior to traditional autoregressive integrated moving average (ARIMA) models and two recently developed neural network models.