# Package com.numericalmethod.suanshu.stats.regression.linear.glm.modelselection

• Interface Summary
Interface Description
BackwardElimination.Step
ForwardSelection.Step
• Class Summary
Class Description
BackwardElimination
Constructs a GLM model for a set of observations using the backward elimination method.
EliminationByAIC
In each step, a factor is dropped if the resulting model has the least AIC, until no factor removal can result in a model with AIC lower than the current AIC.
EliminationByZValue
In each step, the factor with the least z-value is dropped, until all z-values are greater than the critical value (given by the significance level).
ForwardSelection
Constructs a GLM model for a set of observations using the forward selection method.
GLMModelSelection
Given a set of observations {y, X}, we would like to construct a GLM to explain the data.
SelectionByAIC
In each step, a factor is added if the resulting model has the highest AIC, until no factor addition can result in a model with AIC higher than the current AIC.
SelectionByZValue
In each step, the most significant factor is added, until all remaining factors are insignificant.
• Exception Summary
Exception Description
GLMModelSelection.ModelNotFound
Throw a ModelNotFound exception when fail to construct a model to explain the data.