Design And Development Of GP-Based Data Mining Systems
Initially, function using genetic programming (GP) is investigated through symbolic regression for data mining applications. Various kinds of functions are investigated including function learning tasks as well as Boolean functions learning. The objective of the initial investigation is to review ho...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2003
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Initially, function using genetic programming (GP) is investigated through symbolic regression for data mining applications. Various kinds of functions are investigated including function learning tasks as well as Boolean functions learning. The objective of the initial investigation is to review how GP is applied to function learning tasks. The drawbacks of this method are identified.
Hybrid GP technique based on genetic algorithm-program (GA-P) for function learning tasks with variables and constants is investigated. This hybrid GP technique is further expanded to new hybrid GA SA-p. The new hybrid GA SA-P combining genetic algorithms (GA) and simulated annealing (SA) is proposed for function learning tasks with numeric constants. The convergence bahaviour will be compared with existing GP and genetic algorithm-program (GP-P).
Application of Gp is extended to discover interesting rules among data sets. Given a set of data and appropriate parameter settings, GP is used to discover set of rules that describes the relationships that exist among the data.
Finally, GP is investigated as decision tree classifier for classifying binary and multiclass classification problems. The simulation results will be compared with C4.5 decision tree algorithm. |
|---|