Abstract: Currently, China’s economy is transforming from a stage of high-speed growth to a stage of high-quality development. In the face of fierce market competition, digital transformation has become a common choice for enterprises worldwide to explore new growth opportunities and development models. China’s regional development is uneven, and there are obvious differences in the digitalization process and its driving factors between Eastern and Western enterprises. However, there is a lack of research on the extent of digital transformation in different regions. Thus, based on the data of Chinese listed companies from 2007 to 2021, text mining technology was introduced to analyze intangible assets to achieve a more comprehensive and accurate measurement of the degree of enterprise digital transformation. On this basis, the theory of socio-technical system transformation suitable for nonlinear system analysis was introduced to construct the systematic framework of enterprise digital transformation. Using the Random Forest algorithm and partial dependency plots, the critical key system elements of enterprise digital transformation and the nonlinear relationship involved were identified. Results show that, (1) The Text-Asset analysis proposed in this study is more comprehensive and objective in measuring the level of enterprise digital transformation. (2) The level of digital transformation of Chinese enterprises has been rising year by year, and the increase has accelerated significantly since 2014, and the digital transformation of enterprises has entered a rapid development stage. (3) The regional development of digital transformation of Chinese enterprises is remarkably unbalanced. From the static level, the eastern, central, and western regions are decreasing in order; from the year-on-year growth rate, the eastern region is stabilizing, the central region is decreasing, and the western region is rapidly rising. (4) The key drivers of enterprise digital transformation in different regions are general and specific, and the partial dependency plots of each driver are mostly U-shaped or ladder-shaped. Based on the objective measurement of the level of enterprise digital transformation, the analysis of the spatial and temporal characteristics and driving factors of enterprise digital transformation development is conducive to exploring enterprise digital transformation promotion modes suitable for different regions.

Key words: enterprise digital transformation, socio-technical system change theory, text-asset analysis, random forest, partial dependency plots