Classification is the supervised learning technique of data mining which is used to extract hidden useful knowledge over a large volume of databases by predicting the class values based on the predicting attribute values. Of the various techniques, the most widely talked ones include decision tree, probabilistic model and evolutionary algorithms. Recently, the probabilistic and evolutionary techniques are most worked upon, because of the fact that probabilistic models yields high accuracy when there is no attribute dependency in the existing problem and evolutionary algorithms are used to obtain optimal solution over a large search space very quickly when there is less information known about a problem and problem posses attribute dependency. Though there are tradeoffs in each model still there are scopes to improve upon the existing. The proposed approach improves the evolutionary technique such as genetic algorithm by improving the fitness function parameters. Also, in this we compare the genetic algorithm results with Naïve Bayes algorithm results. For the experimental work we have used the nursery data set taken from the UCI Machine Learning Repository.
B. N, Keshavamurthy; Khan, Asad Mohammed; and Toshniwal, Durga
"Improved Genetic Algorithm Based Classification,"
International Journal of Computer Science and Informatics: Vol. 1
, Article 14.
Available at: https://www.interscience.in/ijcsi/vol1/iss3/14