A model to predict Incurred But Not Neported Claims Reserving(IBNR)is studied in this paper.Double generalized linear models are applied to fit claims numbers data and average claims sizes data,respectively.The mean s...A model to predict Incurred But Not Neported Claims Reserving(IBNR)is studied in this paper.Double generalized linear models are applied to fit claims numbers data and average claims sizes data,respectively.The mean square error of prediction is shown also.The model generalize that of Tweedie’s compound Poisson.Moreover,an example on Swiss Motor Insurance data is exhibited,which is shown more efficient.展开更多
In this paper, we propose a new method that enables us to detect and describe the functional modules in complex networks. Using the proposed method, we can classify the nodes of networks into different modules accordi...In this paper, we propose a new method that enables us to detect and describe the functional modules in complex networks. Using the proposed method, we can classify the nodes of networks into different modules according to their pattern of intra- and extra-module links. We use our method to analyze the modular structures of the ER random networks. We find that different modules of networks have different structure properties, such as the clustering coefficient. Moreover, at the same time, many nodes of networks participate different modules. Remarkably, we find that in the ER random networks, when the probability p is small, different modules or different roles of nodes can be Mentified by different regions in the c-p parameter space.展开更多
文摘A model to predict Incurred But Not Neported Claims Reserving(IBNR)is studied in this paper.Double generalized linear models are applied to fit claims numbers data and average claims sizes data,respectively.The mean square error of prediction is shown also.The model generalize that of Tweedie’s compound Poisson.Moreover,an example on Swiss Motor Insurance data is exhibited,which is shown more efficient.
基金The project supported by the State Key Basic Research Program of China under Grant No. 2006CB705500, National Natural Science Foundation of China under Grant No. 60634010, and the Science and Technology Foundation of Beijing Jiaotong University under Grant No. 2006RC044 and New Century Excellent Talents in University under Grant No. NCEF-06-0074
文摘In this paper, we propose a new method that enables us to detect and describe the functional modules in complex networks. Using the proposed method, we can classify the nodes of networks into different modules according to their pattern of intra- and extra-module links. We use our method to analyze the modular structures of the ER random networks. We find that different modules of networks have different structure properties, such as the clustering coefficient. Moreover, at the same time, many nodes of networks participate different modules. Remarkably, we find that in the ER random networks, when the probability p is small, different modules or different roles of nodes can be Mentified by different regions in the c-p parameter space.
