Deep Learning Based Customer Preferences Analysis in Industry 4.0 Environment

Abstract Customer preferences analysis and modelling using deep learning in edge computing environment are critical to enhance customer relationship management that focus on a dynamically changing market place. Existing forecasting methods work well with often seen linear demand patterns but become less accurate intermittent demands the catering industry. In this paper, we introduce throughput model for both short-term long-term aimed at allowing businesses be highly efficient avoid wastage. Moreover, detailed data collected from business online booking system past three years have been used train verify proposed model. Meanwhile, carefully analyzed seasonal conditions as local or national events (event analysis) could had impact sales. The results compared best performing forecast Xgboost autoregressive moving average (ARMA), they suggest method significantly improves accuracy (up 80%) dishes along reduction associated costs labor allocation.