Tips & Tricks
Choosing a Project
VisiRex vs NeuNet
Extracting Rules
Mining Data
Modeling Data
Using Marks

Choosing Between Neural Network and Inductive Rule Extraction
 It depends on the character of your data and the goals of your project.
 Most projects can be done using either method.
Pros and Cons of Neural Networks
 Pro  Will extract complicated mathematical functions that could involve a combination of input fields.
For example, if your target prediction equals the square root of the sum of the squares of three input fields,
the neural net will have no difficulty modeling this mathematical function.
 Pro  Target prediction may be either numeric values or discreet classes.
 Con  All input fields must be numeric.
 Con  The extracted neural net is a mathematical "black box" that offers no human insight into how the predictions are achieved.
Pros and Cons of Inductive Rule Extraction
 Pro  The extracted rule tree is visually appealing because it helps humans to understand the rules behind the data.
This new understanding is "knowledge discovery".
The extracted rule tree can be used manually without further need for the computer.
 Pro  Accepts input fields as either discreet text classes or as numeric values.
 Con  Target prediction must be a discrete class, not a numeric value.
 Con  Will not discover complicated mathematical functions that involve a combination of input fields.
