The paper describes the classification of relationship rule and its effects judging standards and the realization procedure through computers.The paper fully introduces to the relevant knowledge about relationship rule.
变量选择是统计建模的重要环节,选择合适的变量可以建立结构简单、预测精准的稳健模型。本文在logistic回归下提出了新的双层变量选择惩罚方法——adaptive Sparse Group Lasso(adSGL),其独特之处在于基于变量的分组结构进行筛选,实现...变量选择是统计建模的重要环节,选择合适的变量可以建立结构简单、预测精准的稳健模型。本文在logistic回归下提出了新的双层变量选择惩罚方法——adaptive Sparse Group Lasso(adSGL),其独特之处在于基于变量的分组结构进行筛选,实现了组内和组间双层选择。该方法的优点是对各单个系数和组系数采取不同程度的惩罚,避免了过度惩罚大系数,从而提高了模型的估计和预测精度。求解的难点是惩罚似然函数不是严格凸出的,因此本文基于组坐标下降法求解模型,并建立了调整参数的选取准则。模拟分析表明,对比现有代表性方法 Sparse Group Lasso、Group Lasso及Lasso,adSGL法不仅提高了双层选择精度,而且降低了模型误差。最后,本文将adSGL法应用于信用卡信用评分研究,与logistic回归相比,其具有更高的分类精度和稳健性。展开更多
Theory of industrial organization told us that Industry structure affect the profitability, it will also affect the enterprise’s market strategy and will have indirect effect on profitability through the effect of st...Theory of industrial organization told us that Industry structure affect the profitability, it will also affect the enterprise’s market strategy and will have indirect effect on profitability through the effect of strategy. This article present an empirical research on Chinese life insurance industry with the structure equation model as a tool,the results told us that the industrial structure has positive effects on enterprise’s profitability and market strategy,and market strategy has negative effect on profitability so the industry structure has negative indirect effect on profitability, but the positive effect is more power than the negative effect. This article suggest that Chinese should continue it’s opening policy and advance the cooperation in life insurance industry, and the life insurance companies should be enterprising in product innovation and investing strategy.展开更多
文摘The paper describes the classification of relationship rule and its effects judging standards and the realization procedure through computers.The paper fully introduces to the relevant knowledge about relationship rule.
文摘变量选择是统计建模的重要环节,选择合适的变量可以建立结构简单、预测精准的稳健模型。本文在logistic回归下提出了新的双层变量选择惩罚方法——adaptive Sparse Group Lasso(adSGL),其独特之处在于基于变量的分组结构进行筛选,实现了组内和组间双层选择。该方法的优点是对各单个系数和组系数采取不同程度的惩罚,避免了过度惩罚大系数,从而提高了模型的估计和预测精度。求解的难点是惩罚似然函数不是严格凸出的,因此本文基于组坐标下降法求解模型,并建立了调整参数的选取准则。模拟分析表明,对比现有代表性方法 Sparse Group Lasso、Group Lasso及Lasso,adSGL法不仅提高了双层选择精度,而且降低了模型误差。最后,本文将adSGL法应用于信用卡信用评分研究,与logistic回归相比,其具有更高的分类精度和稳健性。
文摘Theory of industrial organization told us that Industry structure affect the profitability, it will also affect the enterprise’s market strategy and will have indirect effect on profitability through the effect of strategy. This article present an empirical research on Chinese life insurance industry with the structure equation model as a tool,the results told us that the industrial structure has positive effects on enterprise’s profitability and market strategy,and market strategy has negative effect on profitability so the industry structure has negative indirect effect on profitability, but the positive effect is more power than the negative effect. This article suggest that Chinese should continue it’s opening policy and advance the cooperation in life insurance industry, and the life insurance companies should be enterprising in product innovation and investing strategy.
