对创新创业发展水平进行合理预测,才能为其日后的发展提供具有前瞻性及合理性的决策,并打下良好的定量研究基础。为了使预测所得的数据更好的为创新创业服务,提出了LSTM神经网络主导下的预测模型,训练与创新创业发展相关的数据。对反应创新创业发展水平的指标进行预测,并与传统回归模型及BP神经网络模型进行对比后发现LSTM模型的显示效果更好。在此基础上,通过对比陕西省和四川省这两个西部重要省份的创新创业发展情况,能够得出两个省份有着相似的创新创业总体发展水平,但具体发展细节与侧重点上则各有不同。四川在创新创业的发展中拥有着更好更大的基础性投入,而陕西在创新创业的发展中拥有着更高更强的技术产出水平,可以看出陕西创新创业发展的效率要比四川强大,较高的效率弥补了人力物力在投入数量上的不足。

Only reasonable prediction of the development level of innovation and entrepreneurship can provide forward-looking and rational decisions for its future development, and lay a good quantitative research foundation. In order to make the predicted data better serve for innovation and entrepreneurship, a prediction model led by LSTM neural network is proposed to train the data related to innovation and entrepreneurship development. After predicting the indicators reflecting the development level of innovation and entrepreneurship, and comparing with the traditional regression model and BP neural network model, it is found that the LSTM model has a better display effect. On this basis, by comparing the development and trend of innovation and entrepreneurship in Shaanxi and Sichuan, two important western provinces, it can be concluded that the two provinces have similar overall development level of innovation and entrepreneurship, but the specific development details and emphasis are different. Sichuan has better and greater basic input in the development of innovation and entrepreneurship, while Shaanxi has higher and stronger technical output level in the development of innovation and entrepreneurship. It can be seen that the efficiency of innovation and entrepreneurship development in Shaanxi is stronger than That in Sichuan, and the higher efficiency makes up for the shortage of human and material resources in the amount of input.

C93

陕西省教育厅2019年度专项科学研究计划(19JK0342)；陕西省教育厅重点项目(20JZ054)；中国(西安)丝绸之路研究院项目(2019HZ11)