Statistical models are, by design, stochastic models.
The purpose of a statistical model is to estimate the values of
parameters describing a process with uncertain outcomes.
In contrast, deterministic models assume there are no, or more
realistically negligible, statistical uncertainties in the model
structure.
A familiar example of a deterministic model is a model predicting
sunrise and sunset times for your location.
Stochastic models incorporate statistical uncertainty, either
in the model structure itself or in the measurement of data that
enter the model.
A familiar example of a stochastic model is weather forecasting,
where predictions are usually associated with expressions of uncertainty.
Biologists use stochastic models to generate fishing or hunting
harvest forecasts, or trends in population abundance, particularly
for exploited or endangered species.
Stochastic models may be forward or inverse.
Forward models generate outcomes (data) from a specified deterministic
model structure, error specification
and parameter values.
Inverse (or estimation) models challenge data with hypothetical
models in order to estimate parameter values and identify the underlying 'true'
deterministic model structure, under an assumed error
specification.
|