The so-called compartmental analysis offers the theoretical basis of tracer kinetic studies which have found wide practical applications in the medical science as well asbehaviormetrics. The output(observation)of the tracer kinetic process is generally represented by a sum of decaying exponentials: q(t)=n Σ i=lAiexp(-ait), where n represents the number of compartments associated with the process. Thus, the compartmental analysis is a method by which the compartment number as well as other pertinent parameters are determined. One of the widely used methods for the estima of the number of compartments is due to Gardner. In practical application, however, method often suffers from “ error ripples ” in the spectra which are introduced in the process of Fourier conversion of the given output curve using only a finite sequences, and which obscure the true peaks corresponding to the decay constant ai. To suppress the error ripples, we have applied Hanning window function filter. To improve the resolution, we have also applied the digital filter which is designed using the maximum likelihood method and maximum entropy method. The experiments using test functions clearly indicate that the proposed method has advantage over the previous method in the improvement of resolution and suppression of noise.
The objectives of this research are to find latent dimensions of values held by Japanese at present time and to forecast its changes over the coming decade. The entire analysis is based on an assumption that value judgement is made with respect to many factors, whence we aim to derive multidimensional spatial representation of values. Then diversification of values will be expressed as individual differences in the space and shifts in values as changes of the configuration. In this frame of the study multidimensional scaling methods were applied to data obtained from the questionaire conducted on 2000 people in Tokyo. It is found that values are likely to change and diversify slightly but not to turn over.
We have conducted a survey concerning attitudes toward fashion, in which we compared the attitude structures of various groups of people, for example male vs. female, or young vs. old. Some of the main findings are as follows: The most important dimension of the structures of attitude toward fashion can be labeled as “ involvement for fashion ” for all groups of people. Another dimension labeled “ favorable vs. unfavorable attitude toward fashion ” was extracted either as the second or as the third factor. The relation between these two foctors was almost consistent for any of the groups. In other words, while the most involved people turned out to be the most favorable, the least involved people were not the most unfavorable, but rather they were neither favorable nor unfavorable toward fashion in general. These fundamental features were moreor less consistent throughout various groups of people, however, the order of the factors extracted, the contribution of each factor to the total variance, and the relations among factors were subtly different from group to group. Statistical techniques such as factor analysis and quantification theory were also proved be very useful to describe these structures of attitude toward fashion.