Trends in Technological Areas using Patent Analytics: A Comparative Study on the Time Series Forecasting of International Patent Classification

Intellectual Properties like Patents, Utility Models, Industrial Designs, and Copyrights are considered the drivers of today’s modern knowledge-based economy. The number of registered and granted Intellectual Properties is one key performance indicator of a state in advancing science and technology. Thus, analyzing the trends is vital as it sets directions for the state. This study aims to analyze the patent trends in the International Patent Classification by incorporating various classical statistical-based techniques and algorithms in time series forecasting. This leads to the generation of insight into the present and future of registered and granted intellectual properties. Autoregression, Seasonal Autoregressive Integrated Moving-Average and Holt Winter’s Exponential Smoothing are the classical time series forecasting algorithms that were compared using three performance indicators, i.e., Root Mean Square Errors, Mean Absolute Errors, and the Coefficient of Determination. It was concluded that the seasonal autoregressive integrated moving average achieved a high performance of several indicators per section. However, it is vital to consider the behaviour of the present data and the appropriate time series forecasting algorithms for forecasting, thus having it on a case-to-case basis.