The G-RRM model

 

This model is very recently proposed by Chorus 2014. This model generalises the RRM2010 model, and has - just as the µRRM model - beside the RRM2010 model also the linear-additive RUM model as a special case.

 

 

In the G-RRM model the fixed constant of one in the attribute level regret function of the RRM2010 model is replaced by a regret-weight variable denoted γ, see the equation below. γ can be estimated for each attribute seperately, or one can estimate the model using one generic γ.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MATLAB

  • Click here for a bundle of MATLAB codes, which includes code to estimate G-RRM-MNL models

 

BISON BIOGEME

  • Click here for BISON BIOGEME G-RRM estimation code to estimate shopping choice data

 

PYTHON BIOGEME

  • Click here for PYTHON BIOGEME G-RRM estimation code to estimate shopping choice data

 

PANDAS BIOGEME

  • Click here for PANDAS BIOGEME G-RRM estimation code to estimate shopping choice data

Apollo R

  • Click here for Apollo R G-RRM estimation code to estimate shopping choice data.

EXAMPLE DATA FILE

  • Click here to download the example shopping choice data file (see Arentze et al. 2005 for more details on the data)

The figure on the right shows the effect of the size of γ on the attribute level regret function. As can be seen, a γ = 1 results in the Classical RRM model. As γ gradually increases the attribute level regret function becomes less convex. In the special case in which γ = 0, the G-RRM model predicts the same choice behaviour as the linear-additive RUM model.

© 2015 Sander van Cranenburgh