Advanced Analytics
Model Review & Validation



Lorum ipsum dolor est...

Vital Questions

A Quality Assurance Service for Analytics Leaders

We will address these and other vital questions:


  • If the problem has multiple optimal solutions, how does the chosen algorithm affect the variability in the answers?
  • Could data be presented that would make the model infeasible?
  • If the solver is integrated into an application, have all appropriate exceptions been trapped?
  • Were best practices observed to facilitate future model updates?

Predictive Analysis

  • Are training and testing data sets appropriately selected and maintained?
  • How often is the model retrained to re-estimate parameters of the model?
  • How well does the methodology consider the following issues:
    • Inferential uncertainty
      • The model infers from one set of observations to another 
    • Missing data
      • Some data is missing and thus inferred in the calculations 
    • Measurement error
      • Numbers are poorly measured, rounded, and don’t represent complete information
    • Surrogate variables
      • Does the model select alternatives for unavailable variables?