Showing papers on "Model selection published in 1978"

Estimating the Dimension of a Model

Abstract: The problem of selecting one of a number of models of different dimensions is treated by finding its Bayes solution, and evaluating the leading terms of its asymptotic expansion. These terms are a valid large-sample criterion beyond the Bayesian context, since they do not depend on the a priori distribution.

Choice of distribution functions for hydrologic design

Abstract: The problem of selecting a bivariate probability density function (pdf) for the simultaneous water stages at both ends of a confluence reach is considered. This pdf is used to compute the expected flood losses in the design of a levee with minimum expected yearly cost. A case study in Hungary illustrates the methodology throughout the paper. Two model selection procedures are compared: ranking the candidate pdf's by the likelihood of the X2 statistic and ranking them by their sample likelihoods. A composite model consisting of a linear combination of candidate pdfs weighted proportionally to their sample likelihoods is also considered. It is found that the two selection procedures lead to different choices, which in the example represent a significant cost variation. Although the ranking of the distributions reduces the uncertainty by imposing an ordering within the candidate set, a unique pdf does not fully account for the model uncertainty. In this sense the composite model seems a more reasonable choice, especially if the decisions are based on expected values.

Model selection for computer simulation of digital m.o.s.f.e.t. l.s.i. circuits

Abstract: The paper describes a procedure for selecting from a collection of models, for the purpose of m.o.s.f.e.t. circuit simulation, a m.o.s.f.e.t. device model that provides the best trade-off between accuracy and computational efficiency. The selection method employs the concept of the accuracy-efficiency product, and is suited to a situation in which only a few models will be used to represent all the devices in the network. Another selection method is also proposed, which employs a ‘model performance function’ and is appropriate for situations in which each device in a network can be represented by a different model. The results of applying these two selection methods to a particular m.o.s.f.e.t. model library are presented

An application of point kriging in optimum variogram model selection

Abstract: A method of grade estimation,.known as kriging, can be used to estimate a grade value at a specific point within an ore deposit. If the estimated points coincide with the location of a drill hole, then the estimated grade can be compared to. the assayed grade, a known value of a regionalized variable. In this way an error of estimation is determined. Kriging employs a variogram model in determination of estimated grades. By varying the parameters of this model and finding the set which, on the average, yields the smallest estimation error, an optimum model is determined. Hopefully, this optimized model will produce block grade values that are closer to actual production figures than are those from an unoptimized model. To test this hypothesis, mine production figures in the form of blast hole grades from the Pima mine were compared with the estimated block grades. Tests were conducted using four variogram models optimized under different conditions, and the original, unoptimized variogram model. Using standard statistical procedures, it can be demonstrated that every optimized model produces grade estimates closer to the produc­ tion data than the unoptimized model.