Skip to content
dkolobok edited this page Mar 16, 2017 · 1 revision

Welcome to the confest wiki!

The package confest is developed to execute the identification process of ODE models.

To identify a model is to specialize the parameter values by taking into account experimental data.

The package capabilities include:

  • Model fitting. Finds the maximum likelihood estimate for the parameter vector.
  • Confidence estimation. Finds confidence intervals for model parameters and confidence bands for model output. Confidence sets can be constructed via the likelihood ratio test method and the bootstrapping method.
  • Profile likelihood and identifiability analysis. Constructs likelihood profiles for parameters to test the identifiability of the model.
  • Optimal design. Finds the design of experiment that is optimal in the sense of maximizing (minimizing) the design criteria, such as a size of a confidence set.
Clone this wiki locally