A Method for the Simulation of Environmental Data Sets.
OLD DOMINION UNIV NORFOLK VA APPLIED MARINE RESEARCH LAB
Pagination or Media Count:
The major goal of this study was to develop a simulation method which could be applied by environmental scientists who may not have a strong background in distributional theory and, moreover, who may not have ready access to a mainframe computer system e.g. an investigator working on a ship or at a field station. A method was developed which allows the simulation of multivariate data sets without requiring a characterization of the distributional shapes of each of the variables. The method is based upon the concept that most data sets can be approximately normalized by a family of power transformations. Conversely, a matrix of normal deviates produced by a random number generator can be adjusted to appropriate means and standard deviations and back-transformed to simulate the shape of the observed data. The method was successful in simulating data sets displaying a wide range of theoretical distributions as well as real data from an ongoing monitoring program. Keywords Statistical analysis.
- Environmental Health and Safety
- Computer Programming and Software