基于DEA-BP神经网络的效率置信区间预测模型研究

效率预测是这几年比较热门的研究话题，然而随着评价系统的复杂性和不确定性，效率的点预测性能会逐渐降低。基于此，提出DEA-BP神经网络置信区间预测模型。构建非阿基米德无穷小的CCR模型，对系统进行效率评价；构建BPNN的置信区间预测模型，将点预测转化为区间预测；通过PICP、NMPIL、CLC等模型进行区间综合验证。将这三个阶段的模型套用到“一带一路”沿线省市的旅游效率预测中，根据预测结果对各个省市进行效率分类并提出改进建议。由于BPNN置信区间预测模型难以确认最佳模型，该结果仍需改进，但具有一定的借鉴作用。

Research on Efficiency Confidence Interval Prediction Model Based on DEA-BP Neural Network

In recent years, efficiency prediction is a hot topic of research. However, with the complexity and uncertainty of the evaluation system, the point prediction performance of efficiency will gradually decrease. Based on this, a DEA-BP neural network confidence interval prediction model is proposed. Firstly, the non-archimedes infinite CCR model is constructed to evaluate the efficiency of the system. Secondly, the confidence interval prediction model of BPNN is constructed, and the point prediction is transformed into interval prediction. Finally, the interval comprehensive verification is carried out by PICP, NMPIL and CLC models. The three-stage model is applied to the tourism efficiency forecast of provinces and cities along the “Belt and Road”, and the efficiency of each province and city is classified according to the forecast results and suggestions for improvement are proposed. Because the BPNN confidence interval prediction model is difficult to confirm the best model, the results of this paper still need to be improved, but it has certain reference.