Multiple Regression

Some examples

  • Prescribing could be modelled as a function of how the drug performs against certain attributes
      Rx = 20 + 2.65 x 'satisfactorily controls condition'
        + 1.35 x 'does not cause weight gain'
        + 1.30 x 'results in high level of compliance'
        + 0.80 x 'has low incidence of side effects'
        - 1.25 x 'requires periodic monitoring'
        - 1.55 x 'potential for drug interactions'
      A unit rise in performance against 'does not cause weight gain' implies a rise of 1.35 in Rx
  • Customer satisfaction can often be modelled as a function of its 'components' e.g. service and value