PATENT CITATION FORECASTING WITH MACHINE LEARNING TECHNIQUES IN SUPPLY CHAIN TECHNOLOGY MANAGEMENT

Abstract

In today's rapidly evolving technological landscape, innovation in supply chain technologies is essential for sustaining competitive advantage. This study aims to forecast patent citations, which are useful for evaluating the quality and potential impact of patents in supply chain management. Using a dataset of 12,225 patents from lens.org, various machine learning models, including Multiple Linear Regression (MLR), Ridge, Lasso, Artificial Neural Networks (ANN), Support Vector Regression (SVR), Regression Trees (RT), and Random Forest (RF), were applied to predict forward patent citations. Model performance was assessed using RMSE and R2 metrics. Among all the models, RF exhibited the highest accuracy (RMSE = 0.0821, MAE = 0.0135). These findings highlight the effectiveness of machine learning, particularly RF, in identifying high-impact patents. This approach offers valuable insights for researchers and practitioners by providing a data-driven method for assessing technological innovation and patent value in the supply chain domain.