Global Parameter Estimation Research Articles

Development of a Structured Sentencing Simulation

Prison sentencing needs to be comprehensive, adapt to changes over time, and avoid racial, gender, and age disparities. The Structured Sentencing Simulation (sss) addresses these needs through a microsimulation of court and correctional systems, providing parameters to generate the hypothetical impact of changes in sentencing policies. The program design utilizes microsimulation techniques including micro-data input, cell-based submodels, dynamic and static aging, dynamic transition models, discrete time slicing, and disaggregation for output distributions. The quality of the results depends upon the collection and processing of large amounts of data describing an existing population of court sentences, demographic characteristics of those sentenced, cell-based estimates of transition probabilities through the criminal justice system, and global parameter estimates of overall system characteristics. The program output includes 5- and 10-year projections of prison, jail, residential treatment, and community service populations In the output from the model all population counts are broken down by sex, race, age, and other demographic variables enabling policy analysts to momtor the extent to which inmate populations are likely to be free of sentencing disparities The program has been implemented in several state sentencing studies. Aspects of the system have implications for simulations in other social contexts.

Comparison of modal parameter estimation techniques on aircraft structural data

Five global time domain parameter estimation methods, least squares, double least squares, total least squares, correlation fit and Smith least squares were compared using the same set of impulse responses taken from the ground vibration test of an aircraft structure. The behaviour of the modal parameter estimates was investigated for changes in the solution order and number of data points analysed. The techniques produced comparable results for the dominant modes, although there were some differences in the estimates of the other modes. The correlation fit and Smith least squares approaches found equivalent estimates to the other methods while using much smaller computational model orders.

Multiple input orthogonal polynomial parameter estimation

The object of this paper is the development of a so-called global modal parameter estimation technique capable of analysing frequency response functions (FRFs) between several input and response stations simultaneously. The technique analyses the FRFs in their natural domain, the frequency domain. Highly consistent estimates of all modal parameters, including repeated modes, can be obtained. The effect of modes outside the analysis band can also be accounted for by explicitly locating these modes, by including residual terms, or by a combination of both. The use of orthogonal polynomials improves the numerical properties of the calculation procedure. It also reduces the order of the identification problem. All pertinent equations have a size which is proportional to the number of modes in the data, and are independent of the number of response and input stations for which data are analysed simultaneously. As a consequence, the technique lends itself readily to implementation with microcomputers.

A Stochastic Differential Equation Model for the Height Growth of Forest Stands

SUMMARY A brief review of the uses and methods of height-growth prediction in forestry is given in ?2. In ?3 the proposed growth model is presented; it consists of a stochastic differential equation related to the Bertalanffy-Richards growth model, and a measurement-error com- ponent. Explicit expressions and an efficient computational procedure for the likelihood function are obtained. In ?4 a method for the simultaneous maximum likelihood estimation of global and local parameters is outlined. The log likelihood function is maximized by a modified Newton method. The special structure of the problem is exploited in order to handle the very large number of variables involved in the optimization. The approach presented has been success- fully implemented, and some computational experience is reported in ?5.

Maximum Likelihood Estimation in Random Coefficient Models

Abstract Previous Monte Carlo studies examining properties of estimators in random coefficient models have been hindered in part by computational difficulties. In particular, determination of maximum likelihood estimators appears sensitive to the computational algorithm used. In a small Monte Carlo experiment, several distinctly motivated algorithms are examined with respect to accuracy and cost in searching for global and local maximum likelihood parameter estimates. A noncalculus oriented approach offers promise. When compared with other estimators, maximum likelihood estimators, so determined, appear to be statistically relatively efficient.