The frequent occurrences of natural disasters in recent years has led to the importanceof areal statistical modeling of natural disasters in the field of environmental statistics. Forexample, understanding the trend of extreme precipitation using areal statistical modelingenables taking preventive measures to avoid damage from flooding. To analyze such data,we need to consider the tail behavior of distributions, and one such approach is extremevalue analysis.
In this paper, we consider multivariate extreme value modeling to flexibly analyze maximaobserved at plural sites. We focus especially on the spatial domain and adopt max-stableprocesses derived from underlying Gaussian random fields to build multivariate extremevalue models and apply them to annual maximum daily precipitation datasets in Japan.For flexible modeling, the models we propose have a generalized extreme value marginaldistribution with covariates. The results of the analysis revealed the relationship of latitudeand altitude to precipitation in the study region. Confidence intervals for 30-year returnlevel calculated using our model became shorter than those using the single site model formost of the sites. The findings also revealed the existence of a temporal trend of annualmaximum daily precipitation, but climate factors such as temperature and El Ni?no do nothave a significant influence. We also computed the current and 2050’s 30-year return levelpredictions for the region, including the sites with no observation. Finally, we comparedthe proposed method with the regional frequency analysis method, which has been usedcommonly in extreme rainfall data analysis in Japan.
Recently, recurrent neural networks have been widely used in many fields. In the present study, we propose nonlinear analysis using recurrent neural network for leukemia disease data. Bone marrow transplants are a standard treatment for acute leukemia. Recovery following bone marrow transplantation is a multi-state process. We can select the optimum hidden unit based on bootstraping. Outliers are identified by using influential analysis. The significance of recurrent connection in recurrent neural networks is also tested. In order to summarize the measure of goodness-of-fit, the deviance on fitting of the recurrent neural network can be bootstrapped. This article examines predictions of probabilities at some points in multi-state survival models for processing a sequence of covariates values. By using recurrent neural networks, we can predict the conditional probability of surviving for the following short-term (say, six months) during the course of the disease with better accuracy than feed-forward neural networks, partial logistic models or Cox's proportional hazards model.
Japanese statistical quality control activities have been aimed at eliminating false data from the companies since the 1950s. However, purposive alteration of quality data that occurred after 2017 in Japanese representative companies caused the credibility of manufacturing in Japan to be lost seriously, while Japan is being pressed for social reform to a data driven society. In this forum, the author discusses the management of excellence, quality and human resources necessary for data driven society. In particular, he would like to discuss what kind of competencies should be trained for data scientists, what is the economic value that data brings, and what social loss is when the data is altered purposively.