polyfitreg


<code>polyfitreg</code> is a modeling function internally called in the code <code>bdynsys</code>dysymod as well as in <code>bdynsys</code>dysymodbestfitmod. It computes the values for the independed variables, based on polynomial terms and computes all possible regression models with all possible combinations of these independ variables (all possible combinations of polynimal terms) and the changes in the dependent variables which are to be predicted. The function estimates the Betas, computes the Sum Square of Error of all possible models as well as their Log Likelihood and R-squares. It returns internally the Sum Square of Error matrices to <code>bdynsys</code>dysymodbestfit and Betas to <code>bdynsys</code>dysymod.Log Likelihood and R-square are printed.

The package bdynsys for panel/longitudinal data combines methods to model
changes in up to four indicators over times as a function of the indicators
themselves and up to three predictors using ordinary differential equations
(ODEs) with polynomial terms that allow to model complex and nonlinear
effects. A Bayesian model selection approach is implemented. The package
provides also visualisation tools to plot phase portraits of the dynamic
system, showing the complex co-evolution of two indicators over time with the
possibility to highlight trajectories for specified entities (e.g. countries,
individuals). Furthermore the visualisation tools allow for making
predictions of the trajectories of specified entities with respect to the
indicators.

polyfitreg: Least Square Regression with Polynomial Terms als Predictors and Changes as Dependent Variable (Internal Function)

Description

Arguments