高新技术企业创新资金配置风险预警的FOA-SVM模型及实证

为了提高创新资金的运营效率和为企业创新活动提供有效支撑,按照高新技术企业创新资金配置的生态化要求,从资金的来源、使用、偿还、收益分配四个维度进行风险分析并构建风险预警指标,给出创新资金配置的生态目标,将支持向量机二分类问题扩展到多分类并与果蝇优化算法结合构建风险预警的FOA-SVM (fruit fly optimization algorithm and support vector machines)模型,确定警度和警限,选取117家医药制造企业2016年数据进行模型验证并与GA-SVM (genetic algorithm and support vector machines)模型进行比较分析,研究表明,FOA-SVM模型分类准确率高而且对于高风险的误判率低,可以实现对高新技术企业创新资金配置风险的分类预警,为优化创新资金配置、防范与控制创新资金配置风险提供依据.

FOA-SVM model and empirical study on risk early warning of innovation fund allocation of high-tech enterprises

In order to improve the operating efficiency of innovative funds and provide effective support for enterprise innovation activities, and according to ecological requirements of high-tech enterprises innovation fund allocation, this paper makes risk analysis from four dimensions including fund source, use, repayment and income distribution. At the same time, it constructs the risk warning index and gives the ecological goal of innovation fund allocation. This paper extends the classification of SVM (support vector machines) model to multi-classification and combines with the FOA (fruit fly optimization algorithm) model to construct a risk-early-warning FOA-SVM model, and then it confirms the warning level and warning border. At last, this paper selects 117 pharmaceutical manufacturing companies with data in 2016 for model validation and compare FOA-SVM model with GA-SVM (genetic algorithm and support vector machines) model. The research shows that FOA-SVM model has high classification accuracy and low error judge rate for high risk, thus it can realize the classified warning of innovation fund allocation risks for high-tech enterprises, which provides basis for optimizing innovation fund allocation and preventing risks of innovation fund allocation.