Sustainable innovation is an important factor for enterprises to obtain core competitiveness in modern society, and only continuous R&D investment can ensure the smooth progress of enterprise innovation. Therefore, this study uses the post-subsidy data of enterprise R&D investment released by Hangzhou Science and Technology Bureau to explore the influencing factors of R&D investment and its sustainability by Chinese high-tech enterprises. In contrast to previous research which mainly focused on the pre-subsidy, this study focuses on the post-subsidy policy of R&D investment sustainability. Empirical analysis methods, such as cross-sectional linear regression and the propensity-score-matching method, were used to draw the following conclusions: (1) The quantity and sustainability of R&D investment of enterprises are obviously unbalanced among regions. Regions where high-tech enterprises are concentrated have higher levels of R&D investment and sustainability than other regions. (2) Under different scales, there are significant differences in the amount and sustainability of R&D investment. Large enterprises have stronger R&D investment strength, while small enterprises have stronger R&D investment willingness. However, the effect of scale on R&D investment will be reduced by regional factors. (3) The evaluation of high-tech enterprises and the enterprises’ R&D investments affect each other. (4) The sustainability of enterprise R&D investment will be affected by enterprise heterogeneity factors, and in turn, the sustainability of enterprise R&D investment will also affect the amount of enterprise R&D investment. Based on the results, the study provides suggestions for the government to make more targeted policies.

Land-use zoning provides an effective tool for designing and implementing differentiated farmland-protection policies. Despite the exponential increase in research on farmland zoning in recent years, little research has comprehensively explored the supply, demand and relationships of the diverse functions of farmland. In this study, multi-sourced datasets and diverse methods, along with GIS, were combined to spatially evaluate the supply, demand and relationship patterns among the production, ecological and landscape-cultural functions of the farmland of Hangzhou City in China, to construct farmland zoning. The results indicate that high production supply was mostly concentrated in flat plains, whereas highly ecological farmland was frequently observed in mountains. Both urban and rural areas had the capacity to provide aesthetics and recreation. Regarding demand, high values were mainly observed near the downtown area. Additionally, supply-and-demand matching (SDM) and multi-functional coupling and coordination degree (MCCD) were evaluated. Among the four basic zones acquired by SDM analysis, two zones dominated by more than one function were further divided into four sub-regions, according to the MCCD values. Ultimately, six farmland-use zones were determined. By considering the supply, demand and relationships of multiple functions, this farmland-zoning program offers new insights into differentiated farmland protection.

In the process of user churn modeling, due to the imbalance between lost users and retained users, the use of traditional classification models often cannot accurately and comprehensively identify users with churn tendency. To address this issue, it is not sufficient to simply increase the misclassification cost of minority class samples in cost-sensitive methods. This paper proposes using the Focal Loss hard example mining technique to add the class weight α and the focus parameter γ to the cross-entropy loss function of LightGBM. In addition, it emphasizes the identification of customers at risk of churning and raises the cost of misclassification for minority and difficult-to-classify samples. On the basis of the preceding ideas, the FocalLoss_LightGBM model is proposed, along with random forests, SVM, XGBoost, and LightGBM. Empirical analysis based on a dataset of credit card users publicly available on the Kaggle website. The AUC, TPR, and G-mean index values were superior to the existing model, which can effectively improve the accuracy and stability of potential lost users.

Monitoring and mapping agricultural cultural ecosystem services (CES) is essential, especially in areas with a sharp contradiction between agricultural land protection and urban development. Despite research assessing CES increasing exponentially in recent years, our knowledge of the CES of agricultural landscapes is still inadequate. This study used four types of agricultural landscapes in Hangzhou, China, as the study area, analyzed their CES spatial patterns, and explored their societal preferences by integrating the multi-sourced datasets, clustering algorithms, and Maxent model. The results indicated that hot spots of agricultural CES correspond to river valley plains, which were also easily vulnerable to urbanization. Moreover, we found that the CES level of paddy field and dry farmland were higher than tea garden and orchard. Based on the above spatial patterns of supply, demand, and flow of CES, we identified four groups of agricultural land by cluster analysis, distinguishing between significant, unimportant, little used, and potential CES. Further, our results showed that natural and human factors could explain societal preferences. This study can provide a valuable basis for stakeholders to develop balanced strategies by the aforementioned results.

Convergence between emerging technologies and traditional industries has become a crucial strategy for enhancing a technology’s competitiveness. Technical convergence (TC) for smart textiles aims to reveal the convergence of emerging technologies with textile technologies, including the field, structure, and critical technologies of the TC. For the empirical analysis, the technology life cycle (TLC) and network analysis method are utilized to observe the TC of 15,125 patent data for textiles from the Derwent Patent Database. The results indicate the following: (1) after 2021, the TC of smart textiles matured, with the number of patents reaching a peak in 2030. (2) Emerging technologies and textile technologies are inextricably linked. In addition to textile technologies, the primary technical fields involved in smart textiles are electronic engineering, tools design, chemical engineering, and mechanical engineering. Electronic engineering is the most common of these fields, accounting for 29.11%. (3) From a structural perspective, the density, breadth, and depth of the TC continues to expand. (4) Measurement, computer technology, and audio technology will be always essential to the TC, whereas electrical machinery, instrumentation, energy technology, other specialized technologies, and chemical engineering have tremendous growth potential. The findings above have substantial implications for the phenomenon of the TCs that have emerged in emerging technology and traditional industry fields. They can also aid the government in formulating policies that promote the transformation and growth of related industries.