Analysis of Driving Factors for Green Economic Development and Innovation Capability Based on Deep Learning

Abstract

This study employs deep learning methods to analyze the key driving factors influencing the proportion of green patent applications across 30 Chinese provinces from 2010 to 2023. By constructing a multilayer perceptron (MLP) model and integrating SHAP value analysis, the marginal contributions of factors such as economic foundations, innovation investment, digitalization levels, and environmental governance to green patent applications were quantitatively evaluated. The results indicate that chemical oxygen demand (COD) emissions and sulfur dioxide (SO₂) emission intensity are the primary barriers to green technological innovation, whereas R&D investment, new product development expenditures, and digital transformation provide substantial support for green technologies. Furthermore, optimizing the employment structure of the tertiary sector and increasing the average years of education per capita are shown to play a significant role in driving the growth of green patent applications. Finally, this paper proposes policy recommendations, including strengthening pollution control, optimizing economic structures, accelerating the development of green service industries, and enhancing human capital, to provide theoretical support and practical insights for achieving green economic transformation.