Sales Forecasting for Fashion Products Considering Lost Sales

Sales forecasting for new products is significantly important fashion retailer companies because prediction with high accuracy helps the company improve management efficiency and customer satisfaction. The low inventory strategy of stock level in each brick-and-mortar store lead to serious censored demand problems, making difficult. In this regard, a two layers (TLs) model proposed paper predict total sales products. first layer, estimated by linear regression (LR). second are modeled as function not only but also inventory. To solve TLs model, gradient-boosting decision tree method (GBDT) used feature selection. Considering heterogeneity products, mixed k-mean algorithm applied product clustering genetic parameter estimation cluster. tested on real-world data from Singapore company, experimental results show that our better than LR, GBDT, support vector (SVR) artificial neural network (ANN) most cases. Furthermore, indicators built: average conversion rate marginal rate, measure products’ competitiveness explore optimal level, respectively, which provide helpful guidance decision-making industry managers.