Superior
handling
of
variable
interactions:
Scorecards deal with interactions at the aggregate level which leads to loss of information for model building.
Neural Networks are able to deal with complex, non-linear interactions of data variables at the individual data record level.
A
quick
example
Better
handling
of
continuous
variables:
Scorecards divide continuous variables into brackets causing abrupt changes at bracket boundaries.
Neural networks treat continuous variables as truly continuous, resulting in smooth operation throughout the range of values.
A
quick
example
