Journal of the Japanese Society of Computational Statistics
Online ISSN : 1881-1337
Print ISSN : 0915-2350
ISSN-L : 0915-2350
Search
Advanced search
OR
Browse
Search
Volume 13, Issue 1
Displaying 1-5 of 5 articles from this issue
Show all abstracts
    • |<
    • <
    • 1
    • >
    • >|
  • VECTOR REPRESENTATION OF ASYMMETRY IN MULTIDIMENSIONAL SCALING
    Hiroshi Yadohisa, Naoto Niki
    2000 Volume 13 Issue 1 Pages 1-14
    Published: 2000
    Released on J-STAGE: December 09, 2009
    DOIhttps://doi.org/10.5183/jjscs1988.13.1
    JOURNAL FREE ACCESS
    Show abstractHide abstract
    A vector model for representing asymmetry in multidimensional scaling is proposed. Given an asymmetric square matrix of (dis) similarity measures for all pairs of objects in the analysis, and assuming that the locations of objects have already been determined from the symmetric component of the data through some suitable multidimensional scaling model, a set of vectors which represents the latent structure of asymmetry in the data matrix is then determined from the remaining skew-symmetric part of the matrix by the model here in proposed. The discussion is extended to the asymmetric structure of the data, giving indices of several types of asymmetry. A real-data application of the vector model is also described.
    View full abstract
    Download PDF (1293K)
  • A NUMERICAL STUDY OF SOLUTIONS IN LATENT CLASS ANALYSIS WITH TWO CLASSES BY A NEW METHOD
    Tsukio Morita
    2000 Volume 13 Issue 1 Pages 15-24
    Published: 2000
    Released on J-STAGE: December 09, 2009
    DOIhttps://doi.org/10.5183/jjscs1988.13.15
    JOURNAL FREE ACCESS
    Show abstractHide abstract
    This paper shows that estimates of the latent parameters are much improved by using a new method obtained from numerical research. This method contains a new and unknown parameter p, and it intends to improve on the estimates by appropriately giving this p. A criterion for deciding the value of the optimum parameter p is proposed.
    View full abstract
    Download PDF (730K)
  • SENSITIVITY ANALYSIS IN SPATIAL STATISTICS: DETECTING INFLUENTIAL OBSERVATIONS IN SPATIAL PREDICTION
    SeungBae Choi, Yutaka Tanaka
    2000 Volume 13 Issue 1 Pages 25-39
    Published: 2000
    Released on J-STAGE: December 09, 2009
    DOIhttps://doi.org/10.5183/jjscs1988.13.25
    JOURNAL FREE ACCESS
    Show abstractHide abstract
    An important problem in spatial statistics is to predict the unobserved value z(s0) at a specified location s0 based on the information of n observations z(sα), α=1, …, n. It can be achieved in three stages of (1) estimating the variograms, (2) fitting a model to the estimated variograms, and (3) applying the so-called ordinary (or universal) kriging. The present article proposes a method to detect influential observations in variogram estimation, variogram model fitting to the estimated variograms, and spatial prediction using the fitted variogram model. To do this, we derive the influence functions for statistics in the above three stages assuming that the underlying process of the observed spatial data is second-order stationary. A real numerical example is analyzed to show the validity or usefulness of the proposed influence functions. Comparison is made with the influence function derived by Gunst and Hartfield (1997).
    View full abstract
    Download PDF (2185K)
  • GENERALIZED CONSTELLATION GRAPH TRANSFORMATION MODEL FOR PREDICTION
    Osamu Sugano
    2000 Volume 13 Issue 1 Pages 41-48
    Published: 2000
    Released on J-STAGE: December 09, 2009
    DOIhttps://doi.org/10.5183/jjscs1988.13.41
    JOURNAL FREE ACCESS
    Show abstractHide abstract
    The Constellation Graph model (Sugano et al., 1988) and Trigonometric Series transformation model (Sugano, 1994) are proposed by a transformation of the explanatory variable space. However, those models are not linear with respect to parameters. So, the approximate values of the parameters are determined by the use of the Monte Carlo method. In this paper, we determine the parameters by the use of the alternative least square method for the model omitting the restriction of the weights. The model to obtain a predicted value utilizing weighted higher moments is called the Generalized Constellation Graph Transformation Model or GCGTM for short. We describe the method in detail and show some examples of application to air pollution data.
    View full abstract
    Download PDF (696K)
  • LIST OF AUTHORS AND TITLES 14TH ANNUAL MEETING OF THE JAPANESE SOCIETY OF COMPUTATIONAL STATISTICS
    2000 Volume 13 Issue 1 Pages 49-53
    Published: 2000
    Released on J-STAGE: December 09, 2009
    DOIhttps://doi.org/10.5183/jjscs1988.13.49
    JOURNAL FREE ACCESS
    Download PDF (375K)
    • |<
    • <
    • 1
    • >
    • >|
feedback
 Top