Rank the models based on their accuracy evaluation and goodness of fit.

[kan-qi]

Rank the models based on their accuracy evaluation and goodness of fit:

1. check the goodness of fit evaluation results for the compared models:

2. check the estimation accuracy result (cross-validation results):

3. put some code to order the models in terms of the above evaluation results:

The ordering criteria:

Estimation accuracy measures:

'mmre': the smaller the better pred15': the larger the better 'pred25': larger -> better 'pred50': larger -> better "mdmre": smaller -> better "mae": smaller -> better

Goodness of fit measures: R2: The larger the better Eta-squared: larger -> better