Post-Balstm a Bagged Lstm Forecasting Ensemble Embedded with a Postponement Framework to Target the Semiconductor Shortage in the Automotive Industry : An Electronics Manufacturing Services Case Study

In the automotive business environment, with intense competition among firms and with the added element of what is known as the “chips or semiconductor shortage,” adequate demand and supply chain planning is not forthright as semiconductors are essential to fabricating printed circuit boards (PCBs), which in turn are necessary to control numerous computerized requirements of the vehicle. This paper presents a manufacturing postponement framework concerted with an ensemble forecasting method based on deep learning networks to predict highly fluctuating PCB demand data for an electronics manufacturing services (EMS) business that serves the automotive industry. The postponement framework effectively considers the modifications intended by the company to address the shortage. Consequently, this research introduces a forecasting method based on multi-layer LSTM networks using a bagging approach embedded with the postponement setting denominated POST-BaLSM to maximize semiconductor use, minimize waste, and assist in resource planning. POST-BaLSTM automatically fits forecasts in the postponement arrangement by considering multiple LSTM models previously optimized with hyperparameters grid-search. We produce point forecasts and prediction intervals using the models’ estimates of quantile calculations. POST-BaLSTM can capture nonlinear patterns while considering the inherent characteristics of non-stationary data and the postponement assembly. In our case study, the analysis indicates that a two-layered POST-BaLSTM approach can achieve nearly 50% improvement in measuring forecast errors compared to statistical and other machine learning methods