Determinants of Government Support for Enterprises Based on Random Forest Analysis: A Case Study of Western Guangdong Province

Abstract

This paper studies the local enterprises that have received subsidies from Chinese local governments. Based on the 397 enterprises in the western part of Guangdong Province, data from three rounds of applications were collected between 2020 and 2024 for this analysis. The contents of the data set are as follows: type of industry, distribution area, revenue level, number of employees and innovation performance. According to the above indicators, a random forest model was employed in the study to examine the reasons for government subsidy distribution. The Model has an accuracy of 61.72%. Based on the above data, the conditions for support by local governments in China have moved away from mainly focusing on enterprise income to giving more weight to per capital innovation results. Characteristics of the industry and the level of enterprise income are also supporting factors for the evaluation.