Bulletin of the Computational Statistics of Japan

This study attempts to determine the core journals, i. e. the important journals, in social science research based on two indicators: the number of articles and the number of citations received. The former is obtained from the International Bibliography of the Social Sciences (IBSS, 1981-1985), which covers four disciplines-economics, political science, sociology and social and cultural anthropology. The latter is extracted from the Social Science Citation Index (SSCI,1985), which provides an 'impact factor' calculated as the ratio between the number of citations received and the number of citable articles published. The top one-third of journals in the above four disciplines selected according to each indicator are defined as the core journals. These journals are classified into five groups, and include those appearing in either or both of the IBSS and the SSCI. Moreover, the core journals in the social sciences as a whole are found according to their interdisciplinary nature. These journals have to be reviewed by social science researchers before they can apply for library collection management.

View full abstract

Show abstractHide abstract

Mixtures of distributions, in particular the multivariate normal distribution, have been used extensively as models in a wide variety of practical situations where the population may be considered to consist of two or more subpopulations mixed in varying proportions. The problem of decomposing such a mixture of distributions is of considerable interest and utility. In this paper, we consuder the problem of estimating suitable initial values for the EM algorithm in the fitting of normal mixture models to multivariate data. It is given that the likelihood equation often has multiple roots for multivariate mixture models, the selection of initial values is serious problem. Discussion is focused here on the use of clustering methods (i. e. hierarchical type and divisive type) to provide suitable initial structure of data set in this context, and we propose the procedure to obtain a global solution. Examples show how proposed procedure works for actual data sets. Moreover, Monte Carlo numerical experiments are presented which involve to exhibit an effectiveness of proposed procedure, and compare a performance of the conventional clustering methods and the mixture approaches to clustering.

View full abstract

Download PDF (1641K)

An initial-clusters-specified method in cluster analysis is proposed. In usual clustering process, the individuals are treated equally or with weights. As against this, the clustering begins with some specific subsets which are treated like the individuals in the proposed method. The discussion has been focused on three hierarchical clustering methods: group average method, centroid method and Ward method. The efficiency of the proposed method is shown as compared with the other methods, ordinary and object-weighted. And also through the artificial numerical examples and a medical data for the aged, it is shown that the result of the proposed method is available for practical applications.

View full abstract

Given an asymmetric matrix whose entry describes relation between a pair of objects, skew-symmetry analysis (SSA) performs singular value decomposition of the skew-symmetric matrix constructed from the original matrix and then displays the asymmetric relationship in a space spanned by singular vectors. There are certain difficulties of interpretation in the usual applications of SSA. To cope with them, this paper suggests a procedure which utilizes both the coefficient of consistence and dominance matrix for SSA. It serves to detect ordinal structure inherent in the data matrix of asymmetric relations. Illustlative examples are presented by applying the procedure to two sets of data, one is concerned with vertical mobility of occupations and the other with confusion among Morse code signals.

View full abstract

Register with J-STAGE for free!