Databases Reference
In-Depth Information
Table A.1
(Continued )
Modeler/Purpose
Advantages
Limitations and Weaknesses
Linear regression
Simple and quick algorithm
Linear only
Model easy to interpret
No interaction between inputs
Well defined measures of significance
Nominal data having cardinality
10
Polynomial regression
Support for linear and limited curvilinear
Additive model - no interaction between inputs
Simple and quick algorithm
Nominal data limited to cardinality
10
Model easy to interpret
Well defined measures of significance
ANN regression
Can be trained to fit almost any data
Will overfit if not monitored closely
Readily detects interaction effects
between inputs
Structure of model difficult to interpret
No available tests for significance available
May settle in sub-optimal locations
Given random starting location, results may vary
from execution to execution
Nominal data limited to cardinality
10
