Innovating knowledge and information for a firm-level automobile demand forecast system : a machine learning perspective

Accurate demand forecasting is important for automotive manufacturing and sales planning because it allows firms to minimize costs and improve their effectiveness. Based on the limitations of existing literature, this paper seeks to establish a novel machine learning-assisted hybrid-input automobile demand forecast model by focusing on the research gaps in input data, methodology, and the scope of demand forecast. To achieve the research aim, the firm-level forecasting performance of the machine learning algorithms based on the hybrid micro-/firm-level (endogenous) and macro-level (exogenous) factors were analyzed to present the optimal approach. The study collected and used monthly vehicle sales and related firm-level data from South Korea from 2011 to 2020. Linear regression, neural network, random forest, stochastic gradient descent, and ensemble learning were used to build the models and verify significant input features using the RReliefF algorithm. The paper presents significant theoretical and managerial contributions that advance the methodological frameworks in the auto demand forecasting literature based on the ML-assisted hybrid-input model, highlighting less well-known endogenous factors that affect company precision to enhance businesses' practical operations.