Affiliations: [a]
School of Economics and Management, Northeast Agricultural University, Harbin, Heilongjiang province, China
| [b]
School of Economics and Management, Heilongjiang Bayi Agricultural University, Daqing, Heilongjiang province, China
Correspondence:
[*]
Corresponding author. Li Dan, School of Economics and Management, Northeast Agricultural University, Harbin, Heilongjiang province, China. E-mail: [email protected].
Abstract: Estimating the compensation risk of agricultural insurance is a hotspot of current research. The related research mainly focuses on the calculation and simulation of catastrophe risk that agricultural insurance may face. On the whole, the compensation risk of agricultural insurance mainly comes from the agricultural disasters, especially the agro meteorological disasters. Compared with property insurance, the overall compensation rate of agricultural insurance is much higher than that of property insurance, so agricultural insurance belongs to high-risk business operation. In the research, the support vector machine is used as the research technology, and the forecast model corresponding to the insurance market is constructed. At the same time, this paper constructs SVM prediction model and VAR-based SVM prediction model. Finally, the prediction accuracy of the SVM prediction model and the VAR-based SVM prediction model are compared and analyzed. The research shows that the prediction accuracy of VAR-based SVM prediction model is higher, that is, it is easier to draw near-realistic prediction results based on parameter optimization. This paper summarizes the research, puts forward its inadequacies and merits, and provides theoretical reference for subsequent related research.
Keywords: Parameter optimization, machine learning, agricultural insurance, forecast model
