NEW ATTRIBUTE CONSTRUCTION IN MIXED DATASETS USING CLASSIFICATION ALGORITHMS

Classification is a challenging task in data mining technique. The main aim of Classification is to build a classifier based on some cases with some attributes to describe the objects or one attribute to describe the group of the objects. Then group the similar data into number of classifiers and it assigns items in a collection to target categories or classes. Finally classifier is used to predict the group attributes of new cases from the domain based on the values of other attributes. Various classification algorithms have been developed to group data into classifiers. However, those classification algorithms works effectively either on pure numeric data or on pure categorical data and most of them performs poorly on mixed categorical and numerical data types. Previous classification algorithms do not handled outliers perfectly. To overcome those disadvantages this paper represents NCA and CCA algorithms for Numerical and Categorical datasets to improve the performance of classification. Results of these proposed algorithms are compared with existing ones based on parameters such as accuracy, precision and F-Measures.